* new feature: regular expressions (based on posix regex)

* bugfix: based on regex, \labels inside sections/chapters etc
          are now correctly handled (ie, html-anchors above <h1> etc)
* improved layout for pkgdesc
This commit is contained in:
Andreas Meyer 2006-04-11 22:08:32 +00:00
parent 64a4aba329
commit 3e9da694df
11 changed files with 321 additions and 76 deletions

2
.gitattributes vendored
View File

@ -1056,6 +1056,8 @@ Manual_tools/src/latex_converter_config/gif/cc_wp.gif -text svneol=unset#unset
Manual_tools/src/latex_converter_config/gif/cc_wr.gif -text svneol=unset#unset
Manual_tools/src/latex_converter_config/gif/cc_zeta.gif -text svneol=unset#unset
Manual_tools/src/mstring.doc -text
Manual_tools/src/regex++.C -text
Manual_tools/src/regex_dictionary.C -text
Manual_tools/sty/eps_tabs/cc_Class.eps -text
Manual_tools/sty/eps_tabs/cc_Class.pdf -text svneol=unset#unset
Manual_tools/sty/eps_tabs/cc_Concept.eps -text

View File

@ -136,7 +136,9 @@ OBJHTML = lexer.o \
cpp_formatting.o \
input.o \
output.o \
cc_extract_html.o
cc_extract_html.o \
regex++.o \
regex_dictionary.o
OBJIMAGES = cc_extract_images_lexer.C cc_extract_images.o

View File

@ -27,6 +27,8 @@
#include <sstream>
#include <cassert>
#include <regex_dictionary.h>
#include <map>
typedef std::map< string, string, Case_insensitive_string_greater_than > String_map;
@ -1082,9 +1084,9 @@ string // macro
handle_chapter( const string&, string param[], size_t n, size_t opt) {
NParamCheck( 1, 0);
push_current_output();
string chapter_title( param[0]);
crop_string( chapter_title);
remove_separator( chapter_title);
//string chapter_title( param[0]);
//crop_string( chapter_title);
//remove_separator( chapter_title);
string new_main_filename;
string new_main_filepath = macroX( "\\lciInputPath");
@ -1345,6 +1347,32 @@ sorted_map_is_empty( const string&, string param[], size_t n, size_t opt) {
}
/* regular expressions */
string
handle_regex_register( const string&, string param[], size_t n, size_t opt) {
NParamCheck( 2, 0);
regex_register( param[0], param[1] );
return string();
}
string
handle_regex_does_match( const string&, string param[], size_t n, size_t opt) {
NParamCheck( 2, 0);
if( regex_does_match( param[0], param[1] ) )
return string("\\lcTrue");
else
return string("\\lcFalse");
}
string
handle_regex_result( const string&, string param[], size_t n, size_t opt) {
NParamCheck( 1, 0);
unsigned int index = atoi( param[0].c_str() );
return regex_get_submatch(index);
}
// Initialize
// ======================================================================
void init_internal_macros() {
@ -1422,7 +1450,7 @@ void init_internal_macros() {
insertInternalGlobalMacro( "\\lciIfFileExists", if_file_exists, 1);
insertInternalGlobalMacro( "\\lciCopyFile", copy_file, 2);
insertInternalGlobalMacro( "\\lciChapter", handle_chapter, 1);
insertInternalGlobalMacro( "\\lciChapter", handle_chapter, 0);
insertInternalGlobalMacro( "\\lciPopOutput", pop_output, 0);
insertInternalGlobalMacro( "\\lciPushOutput", push_output, 1);
@ -1435,6 +1463,10 @@ void init_internal_macros() {
insertInternalGlobalMacro( "\\lciIncludeOnly", handle_include_only, 1);
insertInternalGlobalMacro( "\\lciIfToBeIncluded", handle_to_be_included, 1);
insertInternalGlobalMacro( "\\lciToHtmlWidth", to_html_width, 1);
insertInternalGlobalMacro( "\\lciRegexRegister", handle_regex_register, 2 );
insertInternalGlobalMacro( "\\lcRegexDoesMatch", handle_regex_does_match, 2 );
insertInternalGlobalMacro( "\\lcRegexResult", handle_regex_result, 1 );
}
// EOF //

View File

@ -32,7 +32,6 @@
\newcommand{\ccAuto}{\lcAuto}
% ###########################################################################
% |
% | o
@ -987,6 +986,12 @@
\newcommand{\lciFlushPkgDescription}[2]{%
\lciFlushPkgDescriptionX{#X1}{#X2}{\ccValueofPkgName}%
}
\newcommand{\lciCheckForLabelInsideHeadingX}[2]{%
\lciCheckForLabelInsideHeading{#XC1}{#2}{\lciLocalVariableLabel}%
\lciLocalVariableLabel%
}
\newcommand{\lciFlushPkgDescriptionX}[3]{%
\lciIfEmpty{#X3}{}{%
\lciIfEmpty{\lciCurrentPartName}{}{%
@ -994,24 +999,20 @@
\gdef\lciCurrentPartName{}%
}%
\lciPushOutput{description}%
\lcRawHtml{<h3>}\ccValueofPkgName\lcRawHtml{</h3>}
\lcRawHtml{<table><tr valign="top"><td }%
%\lciPkgDescriptionIsOddEntry
% {\lcRawHtml{ bgcolor="#C0C0C0" }}
% {}
\lcRawHtml{width="58\%">}
\lcRawHtml{<tr><td colspan="3">}%
\lciCheckForLabelInsideHeadingX{\ccValueofPkgName}{\lciCurrentPackageName}%
\lcRawHtml{<h3>}\lciCurrentPackageName\lcRawHtml{</h3></td></tr>}
%\lcRawHtml{<TABLE>}
\lcRawHtml{<tr valign="top"><td width="75\%">}
\ccValueofPkgSummary
%\lcRawHtml{</td><td width="4\%">}
\lcRawHtml{</td><td }%
%\lciPkgDescriptionIsOddEntry
% {}
% {\lcRawHtml{ bgcolor="#C0C0C0" }}
\lcRawHtml{width="38\%">}
\lcRawHtml{</td><td width="2\%">}
\lcRawHtml{</td><td width="23\%">}
\ccValueofPkgMaturity
\ccValueofPkgDependsOn
\ccValueofPkgLicense
\lcRawHtml{<a href="#C1">User Manual</a> &nbsp; <a href="#C2">Reference Manual</a>}
\lcRawHtml{</td></td></table>}
\lcRawHtml{</td></td>}
%\lcRawHtml{</table>}
\lciPopOutput%
\gdef\ccValueofPkgName{}%
%\lciPkgDescriptionIsOddEntry
@ -1052,6 +1053,7 @@
\lciSortedMapClear{TocReferenceLinks}
}
}
% ___________________________________________________________________________
% ###########################################################################
% | EOF

View File

@ -70,8 +70,7 @@ l% ___________________________________________________________________________
% |#########################################################################|
% Load latex_to_html.sty that provides the basic converter capabilities.
% redefined in latex_to_html.sty
\newcommand{\lcFalse}[2]{#2}
\newcommand{\lcTrue}[2]{#1}
@ -95,6 +94,37 @@ l% ___________________________________________________________________________
\newcommand{\lcStyleFile}[1]{\lciInputStyleFile{\lciConfigPath}{#XC1}}
\newcommand{\lciInputStyleFile}[2]{\lciInclude{#XC1#C2.sty}}
}
% ___________________________________________________________________________
% ###########################################################################
% | Regular Expressions
% ###########################################################################
\newcommand{\lcRegexRegister}[1]{%
\def\lciRegexNameToBeRegistered{#1}%
\lciParseUntilDelimiter{\lciRegexRegisterX}}
\newcommand{\lciRegexRegisterX}[1]{\lciRegexRegisterXX{\lciRegexNameToBeRegistered}{#1}}
\newcommand{\lciRegexRegisterXX}[2]{\lciRegexRegister{#XC1}{#C2}}
\lcRegexRegister{checkforlabel}@(.+)(\\label[^\}]+\})(.*)@
\newcommand{\lciCheckForLabelInsideHeading}[3]{%
\lcRegexDoesMatch{checkforlabel}{#1}{%
% xdef doesnt work, either ..
\gdef#2{\lcRegexResult{1}\lcRegexResult{3}}%
\gdef#3{\lcRegexResult{2}}
}{%
\gdef#2{#1}%
\gdef#3{}%
}%
}%
% Load latex_to_html.sty that provides the basic converter capabilities.
% redefined in latex_to_html.sty
\lcStyleFile{latex_to_html}
% print own release number
@ -196,6 +226,7 @@ l% ___________________________________________________________________________
\lciHtmlFileHeader{Package Description}%
\lciHtmlNavigationTop
\lcRawHtml{<H1>Package Description</H1>}
\lcRawHtml{<table>}
\lciPopOutput%
}{%
% print remaining footnotes at the end of the conversion
@ -211,6 +242,7 @@ l% ___________________________________________________________________________
\lciOpenTmpFile{\lciIndexHeader}\lciIndexFileHeader\lciCloseFile
\lciOpenTmpFile{\lciIndexFooter}\lciIndexFileFooter\lciCloseFile
\lciPushOutput{description}%
\lcRawHtml{</table>}
\lciHtmlNavigationBottom
\lciHtmlFileFooter%
\lciPopOutput%
@ -1087,14 +1119,18 @@ l% ___________________________________________________________________________
\gdef{\lciIfHtmlLinks}{\lcTrue}%
\gdef{\lciIfHtmlIndex}{\lcTrue}%
\refstepcounter{chapter}%
\lciChapter{#1}% does also a PushOutput
\lciCheckForLabelInsideHeading{#1}%
{\lciLocalVariableChapterNameTOC}%
{\lciLocalVariableLabel}%
\lciChapter% does also a PushOutput
% cross linking rules
\lciPushOutput{anchor}%
\lcRawHtml{i Next_chapter_link_}\value{part}\_\value{chapter}%
\lcRawHtml{ <div class="NavNext"><hr> Next chapter: <a href="}%
\lcRawHtmlExpanded{\lciReplaceWithCurrentPathToken}%
\lcRawHtmlExpanded{\lciOutputBasename}%
\lcRawHtml{">}#1\lcRawHtml{</a>
\lcRawHtml{">}\lciLocalVariableChapterNameTOC\lcRawHtml{</a>
}%
\lcRawHtml{i Up_chapter_link_}\value{part}\_\value{chapter}%
\lcRawHtml{ <a href="}%
@ -1111,14 +1147,14 @@ l% ___________________________________________________________________________
% however, setting this to a default of TRUE at the beginning of each
% chapter seems safe.
% file header
\lciHtmlFileHeader{#1}%
\lciHtmlFileHeader{\lciLocalVariableChapterNameTOC}%
\lciHtmlNavigationTop
\lciIfLessExpanded{\lciCountersecnumdepth}{0}%
{%
\lciHtmlChapterToc{\thechapter}{#1}%
\lciHtmlChapterToc{\thechapter}{\lciLocalVariableChapterNameTOC}%
\lciHtmlChapterHeader{#2}%
}{%
\lciHtmlChapterToc{\thechapter}{\thechapter~~~#1}%
\lciHtmlChapterToc{\thechapter}{\thechapter~~~\lciLocalVariableChapterNameTOC}%
\lciHtmlChapterHeader{Chapter \thechapter\\#2}%
}%
}
@ -1150,6 +1186,7 @@ l% ___________________________________________________________________________
\newcommand{\section@om}{%
\lciEndSection
\refstepcounter{section}%
\lciLocalVariableLabel
\lciIfLessExpanded{\lciCountersecnumdepth}{1}{%
\lciHtmlSectionToc{\thesection}{#1}%
\lciHtmlSectionHeader{#2}}%

View File

@ -432,19 +432,26 @@
\newcommand{\lciHtmlPartPkgDesciption}[1]{
\lciPushOutput{description}
\lcRawHtml{
<table width="100\%" border="0" cellspacing="0" cellpadding="0"><tr><td>
<tr><td colspan="3">
<h2 class="PkgDescPart"><br>}#XC1\lcRawHtml{</h1>
</td></tr></table>}
</td></tr>}
\lciPopOutput
}
% Args: <header-text>
\newcommand{\lciHtmlChapterHeader}[1]{\lcRawHtml{
<h1>}#C1\lcRawHtml{</h1>
\newcommand{\lciHtmlChapterHeader}[1]{
\lciCheckForLabelInsideHeading{#C1}%
{\lciLocalVariableChapterName}%
{\lciLocalVariableLabel}%
\lciLocalVariableLabel
\lcRawHtml{<h1>}\lciLocalVariableChapterName\lcRawHtml{</h1>
}}
% Args: <anchor-counter> <toc-entry>: Respect secnumdepth counter.
\newcommand{\lciHtmlChapterToc}[2]{%
%\lciCheckForLabelInsideHeading{#2}%
% {\lciLocalVariableChapterName}%
% {\lciLocalVariableLabel}%
\lciIfEmpty{#C1}{}{\lciIfLessExpanded{\lciCountertocdepth}{0}{}{%
\lcRawHtml{<a name="Chapter_}#C1\lcRawHtml{"></a>}}}
\lciIfLessExpanded{\lciCountertocdepth}{0}{}{%
@ -454,7 +461,7 @@
% \lcRawHtml{##Chapter_}#C1 % omit anchor, we start at the page top
\lcRawHtml{ <tr><td class="TocChapter" colspan="6">
<!-- --------------------------------------------------- -->
<a class="TocLinkChapter" href="}\lcRawHtmlExpanded{\lciOut}%
<a class="TocLinkChapter" href="}\lcRawHtmlExpanded{\lciOut}% \lciLocalVariableChapterName
\lcRawHtml{"
>}#C2\lcRawHtml{</a>
</td></tr>
@ -464,12 +471,20 @@
% Args: <header-text>:
\newcommand{\lciHtmlSectionHeader}[1]{\lcRawHtml{
<h2>}#C1\lcRawHtml{</h2>
}}
\newcommand{\lciHtmlSectionHeader}[1]{%
\lciCheckForLabelInsideHeading{#C1}%
{\lciLocalVariableSectionName}%
{\lciLocalVariableLabel}%
\lciLocalVariableLabel
\lcRawHtml{<h2>}\lciLocalVariableSectionName\lcRawHtml{</h2>
}
}
% Args: <anchor-counter> <toc-entry>: Respect secnumdepth counter.
\newcommand{\lciHtmlSectionToc}[2]{%
\lciCheckForLabelInsideHeading{#C2}%
{\lciLocalVariableSectionName}%
{\lciLocalVariableLabel}%
\lciIfLessExpanded{\lciCountertocdepth}{1}{}{%
\lcRawHtml{<a name="Section_}#C1\lcRawHtml{"></a>}
\edef\lciOut{\lciOutputFilename}%
@ -480,16 +495,16 @@
\lcRawHtml{ <td class="TocSection" colspan="5">
<a class="TocLinkSection" href="}\lcRawHtmlExpanded{\lciOut}%
\lcRawHtml{##Section_}#C1\lcRawHtml{"
>}#C2\lcRawHtml{</a>
>}\lciLocalVariableSectionName\lcRawHtml{</a>
</td>
</tr>
}%
\lciPopOutput}}
% Args: <header-text>:
\newcommand{\lciHtmlReferenceHeader}[1]{\lcRawHtml{
<h2>}#C1\lcRawHtml{</h2>
}}
\newcommand{\lciHtmlReferenceHeader}[1]{
\lcRawHtml{<h2>}#C1\lcRawHtml{</h2>}
}
% Args: <anchor-counter> <toc-entry>: Respect secnumdepth counter.
\newcommand{\lciHtmlReferenceToc}[2]{%
@ -514,13 +529,20 @@
% Args: <anchor-counter> <header-text>: if anchor counter is empty, no
% anchor is defined. Respect secnumdepth counter.
\newcommand{\lciHtmlSubsectionHeader}[2]{%
\lciCheckForLabelInsideHeading{#C2}%
{\lciLocalVariableSubSectionName}%
{\lciLocalVariableLabel}%
\lciLocalVariableLabel
\lciIfEmpty{#C1}{}{\lciIfLessExpanded{\lciCountertocdepth}{2}{}{%
\lcRawHtml{<a name="Subsection_}#C1\lcRawHtml{"></a>}}}\lcRawHtml{
<h3>}#C2\lcRawHtml{</h3>
<h3>}\lciLocalVariableSubSectionName\lcRawHtml{</h3>
}}
% Args: <anchor-counter> <toc-entry>
\newcommand{\lciHtmlSubsectionToc}[2]{%
\lciCheckForLabelInsideHeading{#C2}%
{\lciLocalVariableSubSectionName}%
{\lciLocalVariableLabel}%
\lciIfLessExpanded{\lciCountertocdepth}{2}{}{%
\edef\lciOut{\lciOutputFilename}%
\lciPushOutput{toc}%
@ -529,8 +551,7 @@
}\lciTocIndentWidth\lciTocIndentWidth
\lcRawHtml{ <td class="TocSubsection" colspan="4">
<a class="TocLinkSubsection" href="}\lcRawHtmlExpanded{\lciOut}%
\lcRawHtml{##Subsection_}#C1\lcRawHtml{"
>}#C2\lcRawHtml{</a>
\lcRawHtml{##Subsection_}#C1\lcRawHtml{">}\lciLocalVariableSubSectionName\lcRawHtml{</a>
</td>
</tr>
}%

View File

@ -0,0 +1,72 @@
#include <iostream>
#include <mstring.h>
#include <cassert>
#include "regex++.h"
using namespace std;
const unsigned int maxgroups = 20;
vector<string>
Regular_expression::submatch(maxgroups);
Regular_expression::Regular_expression( const char* rx )
{ init( rx ); }
Regular_expression::Regular_expression( const string& rx )
{ init( rx.c_str() ); }
void
Regular_expression::init( const char* rx ) {
assert( rx != NULL );
if( regcomp( &regex, rx, REG_EXTENDED ) != REG_NOERROR ) {
std::cout << "!! Error: regex compilation error" << std::endl;
exit(1);
}
regmatch = ( regmatch_t* ) malloc ( maxgroups * sizeof ( regmatch_t ) );
assert( regmatch != NULL );
}
Regular_expression::~Regular_expression() {
if( regmatch != NULL ) {
regfree ( &regex );
free ( regmatch );
}
}
bool
Regular_expression::match( const string& s ) {
assert( regmatch != NULL );
if( regexec( &regex, s.c_str(), maxgroups, regmatch, 0 ) == REG_NOERROR ) {
int so, eo;
for ( unsigned int i = 0; i < maxgroups; i++ ) {
so = regmatch[ i ].rm_so;
eo = regmatch[ i ].rm_eo;
if ( so != -1 )
submatch[ i ] = s.substr ( so, eo - so );
else
submatch[ i ] = "";
}
return true;
}
return false;
}
bool
Regular_expression::match( const char *s )
{ return Regular_expression::match( string( s ) ); }
const string&
Regular_expression::get_match( unsigned int num ) {
static string emptystring;
if( num >= maxgroups ) {
std::cerr << "!! warning: accessing bad regex submatch #" << num << std::endl;
return emptystring;
}
return submatch[num];
}

View File

@ -0,0 +1,37 @@
#ifndef REGEXPP_H
#define REGEXPP_H
#include <sys/types.h>
#include <regex.h>
#include <mstring.h>
#include <vector>
struct Regular_expression {
struct Exception {};
Regular_expression() : regmatch(NULL) {}
Regular_expression( const string& rx );
Regular_expression( const char* rx );
~Regular_expression();
void init( const char *rx );
bool match( const string& s );
bool match( const char* s );
static const string& get_match( unsigned int num );
protected:
static vector <string> submatch;
regex_t regex;
regmatch_t *regmatch;
};
#endif

View File

@ -0,0 +1,30 @@
#include <regex_dictionary.h>
#include <regex++.h>
#include <iostream>
using namespace std;
typedef hash_map< string, Regular_expression > Regex_map;
static Regex_map regex_map;
void
regex_register( const string& name, const string &regex ) {
regex_map[name].init( regex.c_str() );
//std::cerr << "!! Warning: registered name=[" << name << "] regex=[" << regex << "]" << std::endl;
}
bool
regex_does_match( const string& name, const string& text ) {
bool result = regex_map[name].match( text );
std::cerr << "!! checking " << name << " against " << text << " yields: " << result << std::endl;
std::cerr << "match1 : " << regex_get_submatch(1) << std::endl;
std::cerr << "match2 : " << regex_get_submatch(2) << std::endl;
std::cerr << "match3 : " << regex_get_submatch(3) << std::endl;
return result;
}
const string&
regex_get_submatch( unsigned int num ) {
return Regular_expression::get_match( num );
}

View File

@ -0,0 +1,10 @@
#ifndef REGEX_DICTIONARY_H
#define REGEX_DICTIONARY_H
#include <mstring.h>
void regex_register( const string& name, const string &regex );
bool regex_does_match( const string& name, const string& text );
const string& regex_get_submatch( unsigned int num );
#endif