From d4cbd833ea2b1e5f6b7dd39c10a9eed4c57b9df2 Mon Sep 17 00:00:00 2001 From: Karl Berry Date: Wed, 2 Feb 2022 21:41:21 +0000 Subject: texlogsieve (2feb22) git-svn-id: svn://tug.org/texlive/trunk@61859 c570f23f-e606-0410-a88d-b1316a301751 --- Master/texmf-dist/doc/man/man1/texlogsieve.1 | 24 +- .../texmf-dist/doc/man/man1/texlogsieve.man1.pdf | Bin 40761 -> 40946 bytes .../doc/support/texlogsieve/texlogsieve.pdf | Bin 51179 -> 52118 bytes .../doc/support/texlogsieve/texlogsieve.tex | 40 +- Master/texmf-dist/scripts/texlogsieve/texlogsieve | 858 +++++++++++++++------ 5 files changed, 691 insertions(+), 231 deletions(-) (limited to 'Master/texmf-dist') diff --git a/Master/texmf-dist/doc/man/man1/texlogsieve.1 b/Master/texmf-dist/doc/man/man1/texlogsieve.1 index 41095e2374e..a6656bcc0db 100644 --- a/Master/texmf-dist/doc/man/man1/texlogsieve.1 +++ b/Master/texmf-dist/doc/man/man1/texlogsieve.1 @@ -1,4 +1,4 @@ -.TH TEXLOGSIEVE "1" "January 2022" "texlogsieve 1.0.0-beta-2" "User Commands" +.TH TEXLOGSIEVE "1" "February 2022" "texlogsieve 1.0.0-beta-3" "User Commands" .SH NAME @@ -83,6 +83,22 @@ Include/exclude detailed information on under/overfull boxes in the final summary. With \-\-no\-box\-detail, the summary presents only a list of pages and files that had under/overfull boxes (default enabled). +.TP +\fB\-\-ref\-detail\fR, \fB\-\-no\-ref\-detail\fR +Include/exclude detailed information on undefined references in the final +summary. With \-\-no\-ref\-detail, the summary presents only a list of +undefined references, without page numbers and filenames (default enabled). + +.TP +\fB\-\-cite\-detail\fR, \fB\-\-no\-cite\-detail\fR +Include/exclude detailed information on undefined citations in the final +summary. With \-\-no\-cite\-detail, the summary presents only a list of +undefined citations, without page numbers and filenames (default enabled). + +.TP +\fB\-\-summary\-detail\fR, \fB\-\-no\-summary\-detail\fR +Toggle \-\-box\-detail, \-\-ref\-detail, and \-\-cite\-detail at once. + .TP \fB\-\-heartbeat\fR, \fB\-\-no\-heartbeat\fR Enable/disable progress gauge in page-delay mode (default enabled). @@ -124,8 +140,10 @@ characters in the message, including newlines. If needed, you may precede the string with \[lq]////\[rq], in which case you can use lua-style pattern matching (\c .UR https://www.lua.org/pil/20.2.html -.UE ). Note that the string is used verbatim: you need not (and should not) -enclose it in quotes nor escape special characters such as \[lq]\e\[rq]. +.UE ). Note that the string is used verbatim: you may need to enclose it in +quotes or escape special characters such as \[lq]\e\[rq] for the benefit of +the shell, but such quoting and escaping is unnecessary (and harmful) in the +configuration file. .TP \fB\-\-silence\-file\fR=\fI\,FILENAME OR FILE GLOB\/\fR diff --git a/Master/texmf-dist/doc/man/man1/texlogsieve.man1.pdf b/Master/texmf-dist/doc/man/man1/texlogsieve.man1.pdf index 12059aaf046..0f7b7e77472 100644 Binary files a/Master/texmf-dist/doc/man/man1/texlogsieve.man1.pdf and b/Master/texmf-dist/doc/man/man1/texlogsieve.man1.pdf differ diff --git a/Master/texmf-dist/doc/support/texlogsieve/texlogsieve.pdf b/Master/texmf-dist/doc/support/texlogsieve/texlogsieve.pdf index 35a729df5de..bf2f2df5f9f 100644 Binary files a/Master/texmf-dist/doc/support/texlogsieve/texlogsieve.pdf and b/Master/texmf-dist/doc/support/texlogsieve/texlogsieve.pdf differ diff --git a/Master/texmf-dist/doc/support/texlogsieve/texlogsieve.tex b/Master/texmf-dist/doc/support/texlogsieve/texlogsieve.tex index 3608343743f..eb1ad528700 100644 --- a/Master/texmf-dist/doc/support/texlogsieve/texlogsieve.tex +++ b/Master/texmf-dist/doc/support/texlogsieve/texlogsieve.tex @@ -34,6 +34,8 @@ \usepackage{libertinus} \usepackage[scale=.85]{sourcecodepro} +\usepackage{microtype} +\usepackage{metalogo} \RecordChanges @@ -51,12 +53,20 @@ \texttt{-\/-add-[info\textbar warning]-message} from working} \changes{v1.0.0-beta-2}{2022/01/04}{Misc small bugfixes} +\changes{v1.0.0-beta-3}{2022/02/02}{Add options + \texttt{-\/-summary-detail}, + \texttt{-\/-ref-detail}, + \texttt{-\/-cite-detail}} +\changes{v1.0.0-beta-3}{2022/02/02}{Abort on invalid command-line options} +\changes{v1.0.0-beta-3}{2022/02/02}{Detect ``please rerun`` messages and + add them to the summary} +\changes{v1.0.0-beta-3}{2022/02/02}{Fix line unwrapping with \XeTeX} \begin{document} \title{\textsf{texlogsieve}:\thanks{This document -corresponds to \textsf{texlogsieve}~1.0.0-beta-2, -dated~2022-01-04.}\\[.3\baselineskip] +corresponds to \textsf{texlogsieve}~1.0.0-beta-3, +dated~2022-02-02.}\\[.3\baselineskip] {\normalsize(yet another program to)\\[-.6\baselineskip]} {\large filter and summarize \LaTeX\ log files} } @@ -252,6 +262,26 @@ summary. With \texttt{-\/-no-box-detail}, the summary presents only a list of pages and files that had under/overfull boxes (default enabled). \end{description} +\begin{description} +\item[\texttt{-\/-ref-detail}, \texttt{-\/-no-ref-detail}]~\\ +Include/exclude detailed information on undefined references in the final +summary. With \texttt{-\/-no-ref-detail}, the summary presents only a list of +undefined references, without page numbers and filenames (default enabled). +\end{description} + +\begin{description} +\item[\texttt{-\/-cite-detail}, \texttt{-\/-no-cite-detail}]~\\ +Include/exclude detailed information on undefined citations in the final +summary. With \texttt{-\/-no-cite-detail}, the summary presents only a list +of undefined citations, without page numbers and filenames (default enabled). +\end{description} + +\begin{description} +\item[\texttt{-\/-summary-detail}, \texttt{-\/-no-summary-detail}]~\\ +Toggle \texttt{-\/-box-detail}, \texttt{-\/-ref-detail}, and +\texttt{-\/-cite-detail} at once. +\end{description} + \begin{description} \item[\texttt{-\/-heartbeat}, \texttt{-\/-no-heartbeat}]~\\ Enable/disable progress gauge in page-delay mode (default enabled). @@ -297,8 +327,10 @@ messages: space characters in the provided string match any sequence of whitespace characters in the message, including newlines. If needed, you may precede the string with ``////'', in which case you can use lua-style pattern matching (\url{https://www.lua.org/pil/20.2.html}). Note that the -string is used verbatim: you need not (and should not) enclose it in quotes -nor escape special characters such as ``\textbackslash''. +string is used verbatim: you may need to enclose it in quotes or escape +special characters such as ``\textbackslash'' for the benefit of the +shell, but such quoting and escaping is unnecessary (and harmful) in +the configuration file. \end{description} \begin{description} diff --git a/Master/texmf-dist/scripts/texlogsieve/texlogsieve b/Master/texmf-dist/scripts/texlogsieve/texlogsieve index 590b1826339..14674e60aeb 100755 --- a/Master/texmf-dist/scripts/texlogsieve/texlogsieve +++ b/Master/texmf-dist/scripts/texlogsieve/texlogsieve @@ -157,16 +157,24 @@ than max_print_line characters (by default, 79). Since some messages may be output together on a single line, even short messages may be broken across lines. At the same time, there are quite a few ordinary lines that in practice happen to be max_print_line characters long, -which makes detecting line wrapping a real challenge. Also, for no -apparent reason, LuaTeX wraps some lines at max_print_line characters -and others at max_print_line +1 characters. More: I have seen at least -two cases in which TeX "forgot" to wrap a line, and sometimes there is -a blank line between the wrapped line and its continuation line. And -even more! LuaTeX does not break a line in the middle of a multibyte -UTF-8 character. That is obviously a good idea, but it means some lines -may be broken at lengths smaller than max_print_line. While this may -seem rare, it can happen when parts of the document text are included -in the log, as is the case with over/underfull box messages. +which makes detecting line wrapping a real challenge. And, just to +make things more interesting, sometimes there is a blank line between +the wrapped line and its continuation line. + +pdfTeX (and, I suppose, traditional TeX) counts characters as bytes to +choose where to wrap a line. XeTeX, however, counts utf-8 characters +(I do not know whether that's code points or graphemes), so we need to +take that into consideration. + +LuaTeX adds some extra complications: for no apparent reason, it wraps +some lines at max_print_line characters and others at max_print_line +1 +characters. It also sometimes "forgets" to wrap a line. And more! It +does not break a line in the middle of a multibyte UTF-8 character. +That is obviously a good idea, but it counts the line length in bytes, +like pdfTeX, which means some lines may be broken at lengths smaller +than max_print_line. While this may seem rare, it can happen when parts +of the document text are included in the log, as is the case with +over/underfull box messages. So, if at all possible, it is a very good idea to set max_print_line to a really large value (such as 100,000), effectively disabling line @@ -329,9 +337,8 @@ A simple handler: end function exampleHandler:canDoit(position) - local line if position == nil then position = 0 end - line = Lines:get(position) + local line = Lines:get(position) if line == nil then return false, {} end local first, last = string.find(line, self.pattern) @@ -532,6 +539,7 @@ function main(arg) registerHandlers() registerSummaries() convertFilterStringsToPatterns() + detectEngine() while moreData() do if nextHandler == nil then @@ -679,6 +687,14 @@ function initializeGlobals() -- Does the log file have wrapped lines? -- This may be changed by initializeKpse(). badLogFile = true + + -- When we detect one of the many "please rerun LaTeX" + -- messages, this is set to true (used in showSummary) + SHOULD_RERUN_LATEX = false + + -- detectEngine() may set one of these to true + LUATEX = false + XETEX = false end function initializeKpse() @@ -800,6 +816,22 @@ function convertFilterStringsToPatterns() FORCED_CRITICAL = tmp end +function detectEngine() + local line = logfile:read("*line") + local first = string.find(string.lower(line), '^this is lua') + if first ~= nil then + LUATEX = true + else + first = string.find(string.lower(line), '^this is xe') + if first ~= nil then XETEX = true end + end + + local msg = Message:new() + msg.content = line + msg.severity = DEBUG + dispatch(msg) +end + function processCommandLine(args) HEARTBEAT = true PAGE_DELAY = true @@ -811,6 +843,8 @@ function processCommandLine(args) MINLEVEL = WARNING BE_REDUNDANT = false DETAILED_UNDEROVER_SUMMARY = true + DETAILED_REFERENCE_SUMMARY = true + DETAILED_CITATION_SUMMARY = true SILENCE_STRINGS = {} SILENCE_PKGS = {} -- just the package names @@ -835,52 +869,63 @@ Usage: texlogsieve [OPTION]... [INPUT FILE] texlogsieve reads a LaTeX log file (or the standard input), filters out less relevant messages, and displays a summary report. +texlogsieve reads additional options from the texlogsieverc file +if it exists anywhere in the TeX path (for example, in the current +directory). + Options: - --page-delay, --no-page-delay enable/disable grouping - messages by page before display - --summary, --no-summary enable/disable final summary - --only-summary no filtering, only final summary - --shipouts, --no-shipouts enable/disable reporting shipouts - --repetitions, --no-repetitions allow/prevent repeated messages - --be-redundant, --no-be-redundant present/suppress ordinary messages - that will also appear in the summary - --box-detail, --no-box-detail include/exclude full under/overfull - boxes information in the summary - --heartbeat, --no-heartbeat enable/disable progress gauge - -l LEVEL, --minlevel=LEVEL filter out messages with severity - level lower than [LEVEL]. Valid - levels are DEBUG, INFO, WARNING, - CRITICAL, and UNKNOWN - -u, --unwrap-only no filtering or summary, only - unwrap long, wrapped lines - --silence-package=PKGNAME suppress messages from package - PKGNAME; can be used multiple times - --silence-string=EXCERPT suppress messages containing text - EXCERPT; can be used multiple times - --silence-file=FILENAME suppress messages generated during - processing of FILENAME; can be used - multiple times - --semisilence-file=FILENAME similar to --silence-file, but not - recursive - --add-debug-message=MESSAGE add new recognizable debug message - --add-info-message=MESSAGE add new recognizable info message - --add-warning-message=MESSAGE add new recognizable warning message - --add-critical-message=MESSAGE add new recognizable critical message - --set-to-level-debug=EXCERPT reset severity of messages containing - text EXCERPT to DEBUG; can be used - multiple times - --set-to-level-info=EXCERPT reset severity of messages containing - text EXCERPT to INFO; can be used - multiple times - --set-to-level-warning=EXCERPT reset severity of messages containing - text EXCERPT to WARNING; can be used - multiple times - --set-to-level-critical=EXCERPT reset severity of messages containing - text EXCERPT to CRITICAL; can be used - multiple times - -c cfgfile, --config-file=cfgfile read options from config file - -h, --help give this help list - --version print program version]] + --page-delay, --no-page-delay enable/disable grouping + messages by page before display + --summary, --no-summary enable/disable final summary + --only-summary no filtering, only final summary + --shipouts, --no-shipouts enable/disable reporting shipouts + --repetitions, --no-repetitions allow/prevent repeated messages + --be-redundant, --no-be-redundant present/suppress ordinary messages + that will also appear in the summary + --box-detail, --no-box-detail include/exclude full under/overfull + boxes information in the summary + --ref-detail, --no-ref-detail include/exclude full undefined refs + information in the summary + --cite-detail, --no-cite-detail include/exclude full undefined + citations information in the summary + --summary-detail, --no-summary-detail toggle box-detail, ref-detail, and + cite-detail at once + --heartbeat, --no-heartbeat enable/disable progress gauge + -l LEVEL, --minlevel=LEVEL filter out messages with severity + level lower than [LEVEL]. Valid + levels are DEBUG, INFO, WARNING, + CRITICAL, and UNKNOWN + -u, --unwrap-only no filtering or summary, only + unwrap long, wrapped lines + --silence-package=PKGNAME suppress messages from package + PKGNAME; can be used multiple times + --silence-string=EXCERPT suppress messages containing text + EXCERPT; can be used multiple times + --silence-file=FILENAME suppress messages generated during + processing of FILENAME; can be used + multiple times + --semisilence-file=FILENAME similar to --silence-file, but not + recursive + --add-debug-message=MESSAGE add new recognizable DEBUG message + --add-info-message=MESSAGE add new recognizable INFO message + --add-warning-message=MESSAGE add new recognizable WARNING message + --add-critical-message=MESSAGE add new recognizable CRITICAL message + --set-to-level-debug=EXCERPT reset severity of messages containing + text EXCERPT to DEBUG; can be used + multiple times + --set-to-level-info=EXCERPT reset severity of messages containing + text EXCERPT to INFO; can be used + multiple times + --set-to-level-warning=EXCERPT reset severity of messages containing + text EXCERPT to WARNING; can be used + multiple times + --set-to-level-critical=EXCERPT reset severity of messages containing + text EXCERPT to CRITICAL; can be used + multiple times + -c cfgfile, --config-file=cfgfile read options from given config file + in addition to texlogsieverc + -h, --help give this help list + --version print program version]] for _, line in ipairs(linesToTable(msg)) do print(line) end os.exit(0) @@ -888,7 +933,7 @@ Options: --version if vars.version then - print("texlogsieve 1.0.0-beta-2") + print("texlogsieve 1.0.0-beta-3") print("Copyright (C) 2021, 2022 Nelson Lago ") print("License GPLv3+: GNU GPL version 3 or later " .. ".") @@ -918,6 +963,10 @@ Options: vars = processConfigFile(configFile, vars) end + vars.c = nil + vars['config-file'] = nil + + --unwrap-only -- "-u" if vars['unwrap-only'] or vars.u then @@ -931,6 +980,10 @@ Options: MINLEVEL = DEBUG end + vars.u = nil + vars['unwrap-only'] = nil + + --page-delay --no-page-delay --page-delay=true/false @@ -942,6 +995,10 @@ Options: end if vars['page-delay'] then PAGE_DELAY = true end + vars['page-delay'] = nil + vars['no-page-delay'] = nil + + --only-summary if vars['only-summary'] then ONLY_SUMMARY = true end @@ -955,6 +1012,11 @@ Options: end if vars.summary then SHOW_SUMMARY = true end + vars['only-summary'] = nil + vars.summary = nil + vars['no-summary'] = nil + + --no-shipouts --shipouts --shipouts=true/false @@ -963,6 +1025,10 @@ Options: end if vars.shipouts then SHOW_SHIPOUTS = true end + vars.shipouts = nil + vars['no-shipouts'] = nil + + --minlevel -- "-l" local level @@ -975,10 +1041,19 @@ Options: elseif level == "info" then MINLEVEL = INFO elseif level == "warning" then MINLEVEL = WARNING elseif level == "critical" then MINLEVEL = CRITICAL - else MINLEVEL = UNKNOWN + elseif level == "unknown" then MINLEVEL = UNKNOWN + else + print(' texlogsieve: unknown level "' .. level .. '"') + print(' for help, try "texlogsieve --help"') + print() + os.exit(1) end end + vars.l = nil + vars.minlevel = nil + + --no-repetitions --repetitions --repetitions=true/false @@ -989,6 +1064,10 @@ Options: end if vars.repetitions then SILENCE_REPETITIONS = false end + vars.repetitions = nil + vars['no-repetitions'] = nil + + --be-redundant --no-be-redundant --be-redundant=true/false @@ -1000,6 +1079,31 @@ Options: end if vars['be-redundant'] then BE_REDUNDANT = true end + vars['be-redundant'] = nil + vars['no-be-redundant'] = nil + + + --summary-detail + --no-summary-detail + --summary-detail=true/false + if vars['no-summary-detail'] + or vars['summary-detail'] ~= nil + and not vars['summary-detail'] then + + DETAILED_UNDEROVER_SUMMARY = false + DETAILED_REFERENCE_SUMMARY = false + DETAILED_CITATION_SUMMARY = false + end + if vars['summary-detail'] then + DETAILED_UNDEROVER_SUMMARY = true + DETAILED_REFERENCE_SUMMARY = true + DETAILED_CITATION_SUMMARY = true + end + + vars['summary-detail'] = nil + vars['no-summary-detail'] = nil + + --box-detail --no-box-detail --box-detail=true/false @@ -1011,6 +1115,40 @@ Options: end if vars['box-detail'] then DETAILED_UNDEROVER_SUMMARY = true end + vars['box-detail'] = nil + vars['no-box-detail'] = nil + + + --ref-detail + --no-ref-detail + --ref-detail=true/false + if vars['no-ref-detail'] + or vars['ref-detail'] ~= nil + and not vars['ref-detail'] then + + DETAILED_REFERENCE_SUMMARY = false + end + if vars['ref-detail'] then DETAILED_REFERENCE_SUMMARY = true end + + vars['ref-detail'] = nil + vars['no-ref-detail'] = nil + + + --cite-detail + --no-cite-detail + --cite-detail=true/false + if vars['no-cite-detail'] + or vars['cite-detail'] ~= nil + and not vars['cite-detail'] then + + DETAILED_CITATION_SUMMARY = false + end + if vars['cite-detail'] then DETAILED_CITATION_SUMMARY = true end + + vars['cite-detail'] = nil + vars['no-cite-detail'] = nil + + --no-heartbeat --heartbeat --heartbeat=true/false @@ -1019,12 +1157,19 @@ Options: end if vars.heartbeat then HEARTBEAT = true end + vars.heartbeat = nil + vars['no-heartbeat'] = nil + + if vars.filename == nil then logfile = io.stdin else logfile = assert(io.open(vars.filename, "r")) end + vars.filename = nil + + if vars['silence-string'] then SILENCE_STRINGS = vars['silence-string'] end if vars['silence-package'] then SILENCE_PKGS = vars['silence-package'] end @@ -1035,6 +1180,12 @@ Options: if vars['semisilence-file'] then SEMISILENCE_FILES = vars['semisilence-file'] end + vars['silence-string'] = nil + vars['silence-package'] = nil + vars['silence-file'] = nil + vars['semisilence-file'] = nil + + if vars['add-debug-message'] then for _, msg in ipairs(vars['add-debug-message']) do local pat = stringToPattern(msg) @@ -1071,6 +1222,11 @@ Options: end end + vars['add-debug-message'] = nil + vars['add-info-message'] = nil + vars['add-warning-message'] = nil + vars['add-critical-message'] = nil + if vars['set-to-level-debug'] then FORCED_DEBUG = vars['set-to-level-debug'] @@ -1087,6 +1243,24 @@ Options: if vars['set-to-level-critical'] then FORCED_CRITICAL = vars['set-to-level-critical'] end + + vars['set-to-level-debug'] = nil + vars['set-to-level-info'] = nil + vars['set-to-level-warning'] = nil + vars['set-to-level-critical'] = nil + + + local unknown_options = false + for k, v in pairs(vars) do + print(' texlogsieve: unknown option "' .. k .. '"') + unknown_options = true + end + + if unknown_options then + print(' for help, try "texlogsieve --help"') + print() + os.exit(1) + end end function processConfigFile(configFile, currentVars) @@ -1247,7 +1421,21 @@ function adjustSeverity(msg) end end +function checkRerun(msg) + if msg:checkMatch(msg.rerunMessages) then + -- Rerun messages should be silenced when + -- "--no-be-redundant" is in effect + if not BE_REDUNDANT then msg.redundant = true end + return true + end + + return false +end + function processMessage(msg) + -- can't use short-circuit eval here, we need checkRerun() to always execute + SHOULD_RERUN_LATEX = checkRerun(msg) or SHOULD_RERUN_LATEX + adjustSeverity(msg) if ONLY_SUMMARY or PAGE_DELAY then @@ -1347,10 +1535,15 @@ end function showSummary() local thereIsSomething = false - for _, summary in ipairs(summaries) do - if trim(summary:toString()) ~= "" then - thereIsSomething = true - break + + if SHOULD_RERUN_LATEX then + thereIsSomething = true + else + for _, summary in ipairs(summaries) do + if trim(summary:toString()) ~= "" then + thereIsSomething = true + break + end end end @@ -1369,6 +1562,11 @@ function showSummary() print("") end end + + if SHOULD_RERUN_LATEX then + print("** LaTeX says you should rerun **") + print() + end end heartbeat = {} @@ -1472,26 +1670,30 @@ end epilogueHandler = HandlerPrototype:new() +epilogueHandler.beginPatterns = { + -- This appears in the logfile but not on stdout + "^Here is how much", + -- apparently, pdflatex writes this on stdout: + "^%(see the transcript file for additional information%)", + -- while lualatex writes this on stdout: + "^ *%d+ words of node memory still in use:", +} + function epilogueHandler:canDoit(position) - local line if position == nil then position = 0 end - line = Lines:get(position) + local line = Lines:get(position) if line == nil then return false, {} end - -- This appears in the logfile but not on stdout - local _, last = string.find(line, "^Here is how much") - if last == nil then - -- This appears on stdout (and in the log, of course) - _, last = string.find(line, - "^%(see the transcript file for additional information%)") - else - last = string.len(line) + local last + for _, pat in ipairs(self.beginPatterns) do + _, last = string.find(line, pat) + if last ~= nil then break end end if last == nil then return false, {} else - return true, {last = last} + return true, {last = string.len(line)} end end @@ -1554,9 +1756,8 @@ end underOverFullBoxHandler = HandlerPrototype:new() function underOverFullBoxHandler:canDoit(position) - local line if position == nil then position = 0 end - line = Lines:get(position) + local line = Lines:get(position) if line == nil then return false, {} end local basePattern = "^([UO][nv][de][e]?r)full \\(.)box (%b())" @@ -1810,9 +2011,8 @@ end function stringsHandler:canDoitRecursive(patternLines, position, offset, depth) - local line if position == nil then position = 0 end - line = Lines:get(position) + local line = Lines:get(position) if line == nil then return false, {} end -- skip what was processed in a previous iteration/recursion @@ -1910,7 +2110,13 @@ beginningOfLineDebugStringsHandler.patterns = { '^%s*entering extended mode', '^%s*restricted \\write18 enabled%.', '^%s*%%%&%-line parsing enabled%.', - '^%*%*[%w%.]+', -- "**jobname" + + -- Two diferent ways of saying "**jobname": + '^%*%*' .. filepat .. '$', + -- if the jobname does not include the extension, we use the first + -- part of filepat but also excluding the backslash character + '^%*%*[^%%:;,%=%*%?%|%&%$%#%!%@"\\%`\'%<%>%[%]%{%}]+$', + '^\\[^%s=]+=[^%s=]+', -- "\c@chapter=\count174" "^\\openout%d+%s*=%s*`?[^']+'?%.?", @@ -2036,6 +2242,8 @@ anywhereDebugStringsHandler.patterns = { '^%s*%<' .. filepat .. ', id=.- [%d%.]+pt x [%d%.]+pt%>', '^%s*%', -- '^%s*%<' .. filepat .. '%>', -- + + "^%s*`Fixed Point Package', .- %(C%) Michael Mehlich", } @@ -2203,9 +2411,8 @@ genericLatexHandler.patterns = { } function genericLatexHandler:canDoit(position) - local line if position == nil then position = 0 end - line = Lines:get(position) + local line = Lines:get(position) if line == nil then return false, {} end local last, data @@ -2608,9 +2815,8 @@ openParensHandler.loosePattern = "%s*%(" openParensHandler.pattern = openParensHandler.strictPattern function openParensHandler:canDoit(position) - local line if position == nil then position = 0 end - line = Lines:get(position) + local line = Lines:get(position) if line == nil then return false, {} end local first, last = string.find(line, self.pattern) @@ -2620,6 +2826,21 @@ function openParensHandler:canDoit(position) local filename = guessFilename(position) + -- HACK ALERT: if position > 0, we are "looking into the future" + -- trying to figure out whether to unwrap some line. If there is + -- an open parens with no filename here, we should normally return + -- "true". However, if a close parens exists later on the same line, + -- unwrapping is probably a good idea, so we will lie in this case + -- and say we cannot handle the line. This is not a problem: when + -- the time comes to actually process this line, either both the + -- open and close parens will be embedded in a larger, known message + -- (because of the unwrapping) or we will handle them using the + -- "DUMMY" entry in the stack as usual. + if filename == nil and position > 0 then + local first = string.find(line, '%)') + if first ~= nil then return false, {} end + end + return true, {first = first, filename = filename} -- might be nil end @@ -2662,16 +2883,114 @@ closeParensHandler.strictPattern = "^(%s*)%)" closeParensHandler.loosePattern = "%s*%)" closeParensHandler.pattern = closeParensHandler.strictPattern +-- In lookahead, when we say "we can do it" we actually mean "well, +-- we might be able to do it". This is not a problem: it simply +-- causes handleUnrecognizedMessage to not immediately include the +-- close parens in the unrecognized buffer, leaving it in place to +-- be processed at the next iteration. In this next iteration, it +-- will be at the start of the line, allowing us to examine things +-- more carefully. +function closeParensHandler:lookahead() + local line = Lines:get(0) + if line == nil then return false, {} end + + local first = string.find(line, self.loosePattern) + if first == nil then return false, {} end + + return true, {first = first} +end + +-- When position == 0, we just want to know whether there is a close +-- parens character here; it is up to doit() to match it or not with +-- a file or a DUMMY entry in the stack. +-- +-- When position > 0, we are looking into the future to determine +-- whether we should unwrap a long line. In this case, we may want +-- to say "no, we cannot handle this" even if there is a close +-- parens character here: +-- +-- * The close parens may pair up with something in the stack. If +-- that is the case, we do not want to unwrap a line, as it is +-- an independent message, so we should return true. +-- +-- * The close parens may not pair up with anything. If that is the +-- case, we cannot really know whether we should unwrap or not: +-- it may be the continuation of a previous unknown message or +-- the start of a new unknown message. However, it is unlikely +-- for an unknown message to start with a close parens character, +-- so it's probably better to unwrap and, therefore, we should +-- return false (i.e., lie). Note, however, that this case is +-- very unlikely to happen in practice with parens (it may happen +-- with square brackets): there is always something in the stack, +-- even if it is only the main tex file we are processing. +-- +-- * The close parens may pair up with some open parens character +-- in a line between 0 and the current value of "position". If +-- that is the case, we want to unwrap, as they are probably part +-- of the same message, so we should return false (i.e., lie). +-- +-- * A special case happens if the line we want to decide on whether +-- to unwrap or not is an "open file" message. If that is the case, +-- proceeding to unwrap as per the previous item is "wrong". Still, +-- this causes no harm: during processing of the "open file" +-- message, the close parens will be detected and postponed for +-- future processing. + function closeParensHandler:canDoit(position) - local line if position == nil then position = 0 end - line = Lines:get(position) + local line = Lines:get(position) if line == nil then return false, {} end local first = string.find(line, self.pattern) if first == nil then return false, {} end - return true, {first = first} + -- Ok, there is a close parens character here. We are either at the + -- current line or in the "future" (position > 0). + + if position == 0 then return true, {first = first} end + + -- If we are in the "future", we check for open/close parens characters + -- in the lines between "present" and "future" trying to pair them up, + -- either among themselves or with whatever is at the stack. + local linenum = 0 + local pending = openFiles:size() + local unpaired = 0 -- open parens with no corresponding close parens + + -- This is "<", not "<=". Why? Because the line pointed at by "position" + -- starts at the close parens character, so there is nothing before it + -- to check: If position > 0, this is only called if the close parens + -- character really is at the beginning of the line + while linenum < position do + local size = 0 + local i = 1 + local line = Lines:get(linenum) + if line ~= nil then size = string.len(line) end + + while i <= size do + local j = string.find(line, '[%(%)]', i) + if j ~= nil then + local open = string.find(line, '%(', i) + if open then + unpaired = unpaired +1 + elseif unpaired > 0 then + unpaired = unpaired -1 + elseif pending > 0 then + pending = pending -1 + end + i = j +1 + else + i = size +1 + end + end + + linenum = linenum +1 + end + + if pending > 0 and unpaired == 0 then + return true, {first = first} -- we pair up with something in the stack + else + return false, {} -- let's lie! + end end function closeParensHandler:doit() @@ -2706,9 +3025,8 @@ openSquareBracketHandler.loosePattern = "%s*%[" openSquareBracketHandler.pattern = openSquareBracketHandler.strictPattern function openSquareBracketHandler:canDoit(position) - local line if position == nil then position = 0 end - line = Lines:get(position) + local line = Lines:get(position) if line == nil then return false, {} end local first, last = string.find(line, self.pattern) @@ -2718,6 +3036,12 @@ function openSquareBracketHandler:canDoit(position) local latexPage = guessShipoutPage(position) + -- See the comment "HACK ALERT" in openParensHandler:canDoit() + if latexPage == nil and position > 0 then + local first = string.find(line, '%]') + if first ~= nil then return false, {} end + end + return true, {first = first, latexPage = latexPage} -- may be nil end @@ -2759,16 +3083,63 @@ closeSquareBracketHandler.strictPattern = "^(%s*)%]" closeSquareBracketHandler.loosePattern = "%s*%]" closeSquareBracketHandler.pattern = closeSquareBracketHandler.strictPattern +-- Read the comment right before "closeParensHandler:lookahead()" +function closeSquareBracketHandler:lookahead() + local line = Lines:get(0) + if line == nil then return false, {} end + + local first = string.find(line, self.loosePattern) + if first == nil then return false, {} end + + return true, {first = first} +end + +-- Read the comments right before and inside "closeParensHandler:canDoit()" function closeSquareBracketHandler:canDoit(position) - local line if position == nil then position = 0 end - line = Lines:get(position) + local line = Lines:get(position) if line == nil then return false, {} end local first = string.find(line, self.pattern) if first == nil then return false, {} end - return true, {first = first} + if position == 0 then return true, {first = first} end + + local linenum = 0 + local pending = openFiles:size() + local unpaired = 0 + + while linenum < position do + local size = 0 + local i = 1 + local line = Lines:get(linenum) + if line ~= nil then size = string.len(line) end + + while i <= size do + local j = string.find(line, '[%[%]]', i) + if j ~= nil then + local open = string.find(line, '%[', i) + if open then + unpaired = unpaired +1 + elseif unpaired > 0 then + unpaired = unpaired -1 + elseif pending > 0 then + pending = pending -1 + end + i = j +1 + else + i = size +1 + end + end + + linenum = linenum +1 + end + + if pending > 0 and unpaired == 0 then + return true, {first = first} + else + return false, {} + end end function closeSquareBracketHandler:doit() @@ -2835,9 +3206,8 @@ utf8FontMapHandler = HandlerPrototype:new() -- we repeat here the tests we make on the other methods of this -- object, which is somewhat dirty. function utf8FontMapHandler:canDoit(position) - local line if position == nil then position = 0 end - line = Lines:get(position) + local line = Lines:get(position) if line == nil then return false, {} end local first, encoding @@ -3007,7 +3377,7 @@ function Message:toString(noSilence) if self.severity < MINLEVEL then self.formatted = "" return "" end - if self:redundant() then self.formatted = "" return "" end + if self:ignoreAsRedundant() then self.formatted = "" return "" end self.formatted = self:realToString() if trim(self.formatted) == "" then self.formatted = "" return "" end @@ -3038,10 +3408,83 @@ function Message:realToString() return msg end -function Message:redundant() +Message.redundantMessages = { + { + WARNING, + 'LaTeX', + 'There were undefined references%.' + }, + { + WARNING, + 'LaTeX', + 'There were multiply%-defined labels%.' + }, +} + +Message.rerunMessages = { + { + WARNING, + 'LaTeX', + 'Label%(s%) may have changed%. Rerun to get cross%-references right%.' + }, + { + WARNING, + 'longtable', + 'Table widths have changed%. Rerun LaTeX%.' + }, + { + WARNING, + 'rerunfilecheck', + "File %b`' has changed%." + }, + { + WARNING, + 'biblatex', + 'Please rerun LaTeX%.' + }, +} + +function Message:checkMatch(patlist) + for _, pat in ipairs(patlist) do + -- lua does not have "continue", so we put the loop body + -- in a "repeat/until true" block and use break instead. + repeat + local severity = pat[1] + local pkgname = pat[2] + local text = pat[3] + + if self.severity ~= severity then break end + + -- This code targets messages generated by genericLatexHandler. + -- With it, messages generated by LaTeX do not carry the name + -- of any package; in these cases, we use "LaTeX" instead. + local name = self.name + if name == nil then name = self.what end + if name ~= pkgname then break end + + local first = string.find(self:realToString(), text) + if first ~= nil then return true end + until true + end + return false end +function Message:ignoreAsRedundant() + if BE_REDUNDANT then return false end + + -- this may also be set by checkRerun() + if self.redundant == nil then + if self:checkMatch(self.redundantMessages) then + self.redundant = true + else + self.redundant = false + end + end + + return self.redundant +end + function Message:toSummary() local formatted = self:toString() if trim(formatted) == "" then return end @@ -3139,7 +3582,7 @@ end underOverMessage = Message:new() underOverMessage.severity = WARNING -function underOverMessage:redundant() +function underOverMessage:ignoreAsRedundant() return not BE_REDUNDANT end @@ -3165,7 +3608,7 @@ end missingCharMessage = Message:new() -function missingCharMessage:redundant() +function missingCharMessage:ignoreAsRedundant() return not BE_REDUNDANT end @@ -3187,7 +3630,7 @@ end citationMessage = Message:new() -function citationMessage:redundant() +function citationMessage:ignoreAsRedundant() return not BE_REDUNDANT end @@ -3248,6 +3691,11 @@ function SummaryPrototype:alreadySeen(msg) return self.messages[formatted] ~= nil end +-- we use this for --no-ref-detail and --no-cite-detail +function SummaryPrototype:showDetails() + return true +end + function SummaryPrototype:toString() -- check if the table is empty - https://stackoverflow.com/a/1252776 if next(self.messages) == nil then return "" end @@ -3256,6 +3704,14 @@ function SummaryPrototype:toString() if text == "" then return "" end -- happens with repetitionsSummary + if self.header ~= "" then + if self:showDetails() then + self.header = self.header .. '\n' + else + self.header = self.header .. ' ' + end + end + return self.header .. text end @@ -3270,11 +3726,15 @@ function SummaryPrototype:processAllMessages() for _, messagesSublist in pairsSortedByKeys(self.messages) do local tmp = self:processSingleMessageList(messagesSublist) if tmp ~= "" then - allText = allText .. '\n\n' .. tmp + if self:showDetails() then + allText = allText .. '\n\n' .. tmp + else + allText = allText .. ", " .. tmp + end end end - -- remove leading '\n\n' + -- remove leading '\n\n' or ', ' return string.sub(allText, 3) end @@ -3316,31 +3776,15 @@ function SummaryPrototype:pageAndFileList(messages) files = tmp table.sort(files) - -- Now turn these into strings - tmp = "" - for _, page in ipairs(pages) do - tmp = tmp .. ", " .. page - end - pages = tmp - - local _, last = string.find(pages, '^, ') - if last ~= nil then pages = string.sub(pages, last +1) end - - local tmp = "" - for _, file in ipairs(files) do - tmp = tmp .. ", " .. file - end - files = tmp - - local _, last = string.find(files, '^, ') - if last ~= nil then files = string.sub(files, last +1) end + pages = listToCommaSeparatedString(pages) + files = listToCommaSeparatedString(files) return pages, files end repetitionsSummary = SummaryPrototype:new() -repetitionsSummary.header = 'Repeated messages:\n' +repetitionsSummary.header = 'Repeated messages:' function repetitionsSummary:toString() if not SILENCE_REPETITIONS then return "" end @@ -3376,7 +3820,7 @@ end missingCharSummary = SummaryPrototype:new() -missingCharSummary.header = 'Missing characters:\n' +missingCharSummary.header = 'Missing characters:' function missingCharSummary:processSingleMessageList(messages) local text = "" @@ -3392,7 +3836,11 @@ end citationsSummary = SummaryPrototype:new() -citationsSummary.header = 'Undefined citations:\n' +citationsSummary.header = 'Undefined citations:' + +function citationsSummary:showDetails() + return DETAILED_CITATION_SUMMARY +end function citationsSummary:add(msg) -- group messages by problem key. We do not use msg:toString() @@ -3421,19 +3869,32 @@ function citationsSummary:processSingleMessageList(messages) local key = messages[1].key if key == "" then key = '???' end - text = key .. '\n' + if self:showDetails() then + text = key .. '\n' .. 'in pages ' .. pages .. " (files " .. files .. ")" + else + text = key + end return text end referencesSummary = citationsSummary:new() -referencesSummary.header = 'Undefined references:\n' +referencesSummary.header = 'Undefined references:' + +function referencesSummary:showDetails() + return DETAILED_REFERENCE_SUMMARY +end labelsSummary = citationsSummary:new() -labelsSummary.header = 'Multiply defined labels:\n' +labelsSummary.header = 'Multiply defined labels:' + +-- LaTeX does not supply details for these +function labelsSummary:showDetails() + return false +end -- This is a little different from the others; we do not want to @@ -3554,6 +4015,18 @@ function pairsSortedByKeys (t, f) return iter end +function listToCommaSeparatedString(list) + local tmp = "" + for _, item in ipairs(list) do + tmp = tmp .. ", " .. item + end + + local _, last = string.find(tmp, '^, ') + if last ~= nil then tmp = string.sub(tmp, last +1) end + + return tmp +end + --[[ ##### STACK ##### ]]-- @@ -3872,9 +4345,14 @@ function Lines:gotoNextLine() -- When unwrapping lines, we need to check whether a line is of -- the "right" size. However, we modify the content of currentLine - -- during processing, so we capture its initial length here + -- during processing, so we capture its initial length here. See + -- Lines:wrappingLength(). if self.current ~= nil then - self.currentLineInitialLength = string.len(self.current) + if XETEX then + self.currentLineInitialLength = utf8.len(self.current) + else + self.currentLineInitialLength = string.len(self.current) + end else self.currentLineInitialLength = 0 self.currentWrapped = false @@ -3898,7 +4376,11 @@ end function Lines:len(n) local n = n or 0 if n == 0 then return self.currentLineInitialLength end - return string.len(self[n]) + if XETEX then + return utf8.len(self[n]) + else + return string.len(self[n]) + end end function Lines:append(x) @@ -3981,20 +4463,32 @@ end function Lines:wrappingLength(position) local line = self:get(position) - local n = self:len(position) + local n = self:len(position) -- with XeTeX, this uses utf8.len() + + -- pdfTeX and XeTeX simply wrap at max_print_line (default 79): + -- pdfTeX counts bytes and XeTeX counts utf8 chars. I do not know + -- whether "chars" here means "code points" or "graphemes"; lua's + -- utf8.len() uses code points, which is probably good enough. + if not LUATEX and n == max_print_line then return true end + + -- With LuaTeX, we need to handle a few special cases. - -- I have seen at least two cases where TeX "forgot" to - -- wrap a line. In this happens, the line is not wrapped. + -- LuaTeX sometimes "forgets" to wrap a line. If this happens, the + -- line is not wrapped at all. Why do we do max_print_line +1 here? + -- Because LuaTeX sometimes wraps at max_print_line and sometimes + -- at max_print_line +1. if n > max_print_line +1 then return false end -- Is the line the "right" length? -- (max_print_line or max_print_line +1) if n >= max_print_line then return true end - -- Ok, n < max_print_line, so it looks like this is not - -- a wrapped line. BUT! LuaTeX does not break a multibyte - -- UTF-8 character, which means some lines may be broken - -- at lengths < max_print_line; let's check for this. + -- Ok, n < max_print_line, so it looks like this is not a wrapped + -- line. BUT! LuaTeX tries to wrap lines by counting bytes just as + -- pdfTeX (and differently from XeTeX). However, it does not break + -- a multibyte UTF-8 character (which is obviously good). This means + -- some lines may be broken at lengths < max_print_line; let's check + -- for this. -- Get the length of the first UTF-8 character on the next line: -- https://www.lua.org/manual/5.3/manual.html#pdf-utf8.charpattern @@ -4019,103 +4513,19 @@ function Lines:wrappingLength(position) end function Lines:noHandlersForNextLine(position) - local line = self:get(position) - - local wrapped = true for _, candidateHandler in ipairs(beginningOfLineHandlers) do - if candidateHandler:canDoit(position +1) then wrapped = false break end + if candidateHandler:canDoit(position +1) then + return false + end end for _, candidateHandler in ipairs(anywhereHandlers) do - if candidateHandler:canDoit(position +1) then wrapped = false break end - end - - if wrapped then return true end - - -- TODO: the way we deal with these special cases is lame... - - -- a close parens at the beginning of the next line might not be - -- a "close file" message, but instead refer to an open parens in - -- the current line; let's check for this exceptional case. We - -- traverse the current line looking for open/close parens and pair - -- every close parens to the corresponding open parens. A dangling - -- close parens is probably a continuation too, so we ignore it. - if closeParensHandler:canDoit(position +1) then - -- We previously decided the line is not wrapped because - -- there is a close parens in the following line; let's - -- revise that decision - local unpaired = 0 - local i = 1 - local size = string.len(line) - while i <= size do - local j = string.find(line, '[%(%)]', i) - if j ~= nil then - local open = string.find(line, '%(', i) - if open then - unpaired = unpaired +1 - elseif unpaired > 0 then - unpaired = unpaired -1 - end - i = j +1 - else - i = size +1 - end - end - - if unpaired > 0 then return true end - end - - -- Same thing for close square bracket - if closeSquareBracketHandler:canDoit(position +1) then - -- We previously decided the line is not wrapped because - -- there is a close square bracket in the following line; - -- let's revise that decision - local unpaired = 0 - local i = 1 - local size = string.len(line) - while i <= size do - local j = string.find(line, '[%[%]]', i) - if j ~= nil then - local open = string.find(line, '%[', i) - if open then - unpaired = unpaired +1 - elseif unpaired > 0 then - unpaired = unpaired -1 - end - i = j +1 - else - i = size +1 - end + if candidateHandler:canDoit(position +1) then + return false end - - if unpaired > 0 then return true end end - local parens, data = openParensHandler:canDoit(position +1) - local filename = data[1] - if parens and filename == nil then - -- Next line could be handled as an "open parens" message, but - -- there is no filename, so the message is only "(". There is - -- a good chance that is just the continuation of a previous - -- wrapped message. If there is a close parens character on the - -- line too, we'll assume that is true and ignore the handler - local first = string.find(self:get(position +1), "%)") - if first ~= nil then return true end - end - - local bracket, data = openSquareBracketHandler:canDoit(position +1) - local page = data[1] - if bracket and page == nil then - -- Next line could be handled as an "open square bracket" message, - -- but there is no page number, so the message is only "[". There - -- is a good chance that is just the continuation of a previous - -- wrapped message. If there is a close square bracket character on - -- the line too, we'll assume that is true and ignore the handler - local first = string.find(self:get(position +1), "%]") - if first ~= nil then return true end - end - - return false + return true end function unwrapUntilPatternMatches(pat) -- cgit v1.2.3