1: #!/usr/bin/perl
2:
3: # Scott Harrison
4: # YEAR=2001
5: # May 2001
6: # 06/19/2001,06/20,06/24 - Scott Harrison
7: # 9/5/2001,9/6,9/7,9/8 - Scott Harrison
8: # 9/17,9/18 - Scott Harrison
9: # 11/4,11/5,11/6,11/7,11/16,11/17 - Scott Harrison
10: #
11: # $Id: lpml_parse.pl,v 1.26 2001/12/05 01:39:08 harris41 Exp $
12: ###
13:
14: ###############################################################################
15: ## ##
16: ## ORGANIZATION OF THIS PERL SCRIPT ##
17: ## 1. Notes ##
18: ## 2. Get command line arguments ##
19: ## 3. First pass through (grab distribution-specific information) ##
20: ## 4. Second pass through (parse out what is not necessary) ##
21: ## 5. Third pass through (translate markup according to specified mode) ##
22: ## 6. Functions (most all just format contents of different markup tags) ##
23: ## 7. POD (plain old documentation, CPAN style) ##
24: ## ##
25: ###############################################################################
26:
27: # ----------------------------------------------------------------------- Notes
28: #
29: # I am using a multiple pass-through approach to parsing
30: # the lpml file. This saves memory and makes sure the server
31: # will never be overloaded.
32: #
33: # This is meant to parse files meeting the lpml document type.
34: # See lpml.dtd. LPML=Linux Packaging Markup Language.
35:
36: use HTML::TokeParser;
37:
38: my $usage=<<END;
39: **** ERROR ERROR ERROR ERROR ****
40: Usage is for lpml file to come in through standard input.
41: 1st argument is the mode of parsing.
42: 2nd argument is the category permissions to use (runtime or development)
43: 3rd argument is the distribution (default,redhat6.2,debian2.2,redhat7.1,etc).
44: 4th argument is to manually specify a sourceroot.
45: 5th argument is to manually specify a targetroot.
46:
47: Only the 1st argument is mandatory for the program to run.
48:
49: Example:
50:
51: cat ../../doc/loncapafiles.lpml |\\
52: perl lpml_parse.pl html development default /home/sherbert/loncapa /tmp/install
53: END
54:
55: # ------------------------------------------------- Grab command line arguments
56:
57: my $mode;
58: if (@ARGV==5) {
59: $mode = shift @ARGV;
60: }
61: else {
62: @ARGV=();shift @ARGV;
63: while(<>){} # throw away the input to avoid broken pipes
64: print $usage;
65: exit -1; # exit with error status
66: }
67:
68: my $categorytype;
69: if (@ARGV) {
70: $categorytype = shift @ARGV;
71: }
72:
73: my $dist;
74: if (@ARGV) {
75: $dist = shift @ARGV;
76: }
77:
78: my $targetroot;
79: my $sourceroot;
80: if (@ARGV) {
81: $sourceroot = shift @ARGV;
82: }
83: if (@ARGV) {
84: $targetroot = shift @ARGV;
85: }
86: $sourceroot=~s/\/$//;
87: $targetroot=~s/\/$//;
88:
89: my $logcmd='| tee -a WARNINGS';
90:
91: my $invocation;
92: # --------------------------------------------------- Record program invocation
93: if ($mode eq 'install' or $mode eq 'configinstall' or $mode eq 'build') {
94: $invocation=(<<END);
95: # Invocation: STDINPUT | lpml_parse.pl
96: # 1st argument (mode) is: $mode
97: # 2nd argument (category type) is: $categorytype
98: # 3rd argument (distribution) is: $dist
99: # 4th argument (targetroot) is: described below
100: # 5th argument (sourceroot) is: described below
101: END
102: }
103:
104: # ---------------------------------------------------- Start first pass through
105: my @parsecontents = <>;
106: my $parsestring = join('',@parsecontents);
107: my $outstring;
108:
109: # Need to make a pass through and figure out what defaults are
110: # overrided. Top-down overriding strategy (leaves don't know
111: # about distant leaves).
112:
113: my @hierarchy;
114: $hierarchy[0]=0;
115: my $hloc=0;
116: my $token;
117: $parser = HTML::TokeParser->new(\$parsestring) or
118: die('can\'t create TokeParser object');
119: $parser->xml_mode('1');
120: my %hash;
121: my $key;
122: while ($token = $parser->get_token()) {
123: if ($token->[0] eq 'S') {
124: $hloc++;
125: $hierarchy[$hloc]++;
126: $key=$token->[1].join(',',@hierarchy[0..($hloc-1)]);
127: my $thisdist=' '.$token->[2]{'dist'}.' ';
128: if ($thisdist eq ' default ') {
129: $hash{$key}=1; # there is a default setting for this key
130: }
131: elsif ($dist && $hash{$key}==1 && $thisdist=~/\s$dist\s/) {
132: $hash{$key}=2; # disregard default setting for this key if
133: # there is a directly requested distribution match
134: }
135: }
136: if ($token->[0] eq 'E') {
137: $hloc--;
138: }
139: }
140:
141: # --------------------------------------------------- Start second pass through
142: undef $hloc;
143: undef @hierarchy;
144: undef $parser;
145: $hierarchy[0]=0;
146: $parser = HTML::TokeParser->new(\$parsestring) or
147: die('can\'t create TokeParser object');
148: $parser->xml_mode('1');
149: my $cleanstring;
150: while ($token = $parser->get_token()) {
151: if ($token->[0] eq 'S') {
152: $hloc++;
153: $hierarchy[$hloc]++;
154: $key=$token->[1].join(',',@hierarchy[0..($hloc-1)]);
155: my $thisdist=' '.$token->[2]{'dist'}.' ';
156: # This conditional clause is set up to ignore two sets
157: # of invalid conditions before accepting entry into
158: # the cleanstring.
159: if ($hash{$key}==2 and
160: !($thisdist eq ' ' or $thisdist =~/\s$dist\s/)) {
161: if ($token->[4]!~/\/>$/) {
162: $parser->get_tag('/'.$token->[1]);
163: $hloc--;
164: }
165: }
166: elsif ($thisdist ne ' ' and $thisdist!~/\s$dist\s/ and
167: !($thisdist eq ' default ' and $hash{$key}!=2)) {
168: if ($token->[4]!~/\/>$/) {
169: $parser->get_tag('/'.$token->[1]);
170: $hloc--;
171: }
172: }
173: else {
174: $cleanstring.=$token->[4];
175: }
176: if ($token->[4]=~/\/>$/) {
177: $hloc--;
178: }
179: }
180: if ($token->[0] eq 'E') {
181: $cleanstring.=$token->[2];
182: $hloc--;
183: }
184: if ($token->[0] eq 'T') {
185: $cleanstring.=$token->[1];
186: }
187: }
188: $cleanstring=&trim($cleanstring);
189: $cleanstring=~s/\>\s*\n\s*\</\>\</g;
190:
191: # ---------------------------------------------------- Start final pass through
192:
193: # storage variables
194: my $lpml;
195: my $categories;
196: my $category;
197: my $category_att_name;
198: my $category_att_type;
199: my $chown;
200: my $chmod;
201: my $abbreviation; # space-free abbreviation; esp. for image names
202: my $rpm;
203: my $rpmSummary;
204: my $rpmName;
205: my $rpmVersion;
206: my $rpmRelease;
207: my $rpmVendor;
208: my $rpmBuildRoot;
209: my $rpmCopyright;
210: my $rpmGroup;
211: my $rpmSource;
212: my $rpmAutoReqProv;
213: my $rpmdescription;
214: my $rpmpre;
215: my $directories;
216: my $directory;
217: my $targetdirs;
218: my $targetdir;
219: my $categoryname;
220: my $description;
221: my $files;
222: my $fileglobs;
223: my $links;
224: my $file;
225: my $link;
226: my $fileglob;
227: my $sourcedir;
228: my $targets;
229: my $target;
230: my $source;
231: my $note;
232: my $build;
233: my $buildlink;
234: my $commands;
235: my $command;
236: my $status;
237: my $dependencies;
238: my $dependency;
239: my @links;
240: my %categoryhash;
241: my $dpathlength;
242: my %fab; # file category abbreviation
243:
244: my @buildall;
245: my @buildinfo;
246:
247: my @configall;
248:
249: # Make new parser with distribution specific input
250: undef $parser;
251: $parser = HTML::TokeParser->new(\$cleanstring) or
252: die('can\'t create TokeParser object');
253: $parser->xml_mode('1');
254:
255: # Define handling methods for mode-dependent text rendering
256:
257: $parser->{textify}={
258: targetroot => \&format_targetroot,
259: sourceroot => \&format_sourceroot,
260: categories => \&format_categories,
261: category => \&format_category,
262: abbreviation => \&format_abbreviation,
263: targetdir => \&format_targetdir,
264: chown => \&format_chown,
265: chmod => \&format_chmod,
266: rpm => \&format_rpm,
267: rpmSummary => \&format_rpmSummary,
268: rpmName => \&format_rpmName,
269: rpmVersion => \&format_rpmVersion,
270: rpmRelease => \&format_rpmRelease,
271: rpmVendor => \&format_rpmVendor,
272: rpmBuildRoot => \&format_rpmBuildRoot,
273: rpmCopyright => \&format_rpmCopyright,
274: rpmGroup => \&format_rpmGroup,
275: rpmSource => \&format_rpmSource,
276: rpmAutoReqProv => \&format_rpmAutoReqProv,
277: rpmdescription => \&format_rpmdescription,
278: rpmpre => \&format_rpmpre,
279: directories => \&format_directories,
280: directory => \&format_directory,
281: categoryname => \&format_categoryname,
282: description => \&format_description,
283: files => \&format_files,
284: file => \&format_file,
285: fileglob => \&format_fileglob,
286: links => \&format_links,
287: link => \&format_link,
288: linkto => \&format_linkto,
289: source => \&format_source,
290: target => \&format_target,
291: note => \&format_note,
292: build => \&format_build,
293: status => \&format_status,
294: dependencies => \&format_dependencies,
295: buildlink => \&format_buildlink,
296: glob => \&format_glob,
297: sourcedir => \&format_sourcedir,
298: filenames => \&format_filenames,
299: };
300:
301: my $text;
302: my $token;
303: undef $hloc;
304: undef @hierarchy;
305: my $hloc;
306: my @hierarchy2;
307: while ($token = $parser->get_tag('lpml')) {
308: &format_lpml(@{$token});
309: $text = &trim($parser->get_text('/lpml'));
310: $token = $parser->get_tag('/lpml');
311: print $lpml;
312: print "\n";
313: # $text=~s/\s*\n\s*\n\s*/\n/g;
314: print $text;
315: print "\n";
316: print &end();
317: }
318: exit;
319:
320: # ---------- Functions (most all just format contents of different markup tags)
321:
322: # ------------------------ Final output at end of markup parsing and formatting
323: sub end {
324: if ($mode eq 'html') {
325: return "</body></html>\n";
326: }
327: if ($mode eq 'install') {
328: return '';
329: }
330: }
331:
332: # ----------------------- Take in string to parse and the separation expression
333: sub extract_array {
334: my ($stringtoparse,$sepexp) = @_;
335: my @a=split(/$sepexp/,$stringtoparse);
336: return \@a;
337: }
338:
339: # --------------------------------------------------------- Format lpml section
340: sub format_lpml {
341: my (@tokeninfo)=@_;
342: my $date=`date`; chop $date;
343: if ($mode eq 'html') {
344: $lpml=<<END;
345: <html>
346: <head>
347: <title>LPML Description Page
348: (dist=$dist, categorytype=$categorytype, $date)</title>
349: </head>
350: <body>
351: END
352: $lpml .= "<br /><font size='+2'>LPML Description Page (dist=$dist, ".
353: "categorytype=$categorytype, $date)".
354: "</font>";
355: $lpml .=<<END;
356: <ul>
357: <li><a href='#about'>About this file</a></li>
358: <li><a href='#ownperms'>File Type Ownership and Permissions
359: Descriptions</a></li>
360: <li><a href='#package'>Software Package Description</a></li>
361: <li><a href='#directories'>Directory Structure</a></li>
362: <li><a href='#files'>Files</a></li>
363: </ul>
364: END
365: $lpml .=<<END;
366: <br /> <br /><a name='about' />
367: <font size='+2'>About this file</font>
368: <p>
369: This file is generated dynamically by <tt>lpml_parse.pl</tt> as
370: part of a development compilation process. Author: Scott
371: Harrison (harris41\@msu.edu).
372: </p>
373: END
374: }
375: elsif ($mode eq 'text') {
376: $lpml = "LPML Description Page (dist=$dist, $date)";
377: $lpml .=<<END;
378:
379: * About this file
380: * Software Package Description
381: * Directory Structure
382: * File Type Ownership and Permissions
383: * Files
384: END
385: $lpml .=<<END;
386:
387: About this file
388:
389: This file is generated dynamically by lpml_parse.pl as
390: part of a development compilation process. Author: Scott
391: Harrison (harris41\@msu.edu).
392:
393: END
394: }
395: elsif ($mode eq 'install') {
396: print '# LPML install targets. Linux Packaging Markup Language,';
397: print ' by Scott Harrison 2001'."\n";
398: print '# This file was automatically generated on '.`date`;
399: print "\n".$invocation;
400: $lpml .= "SHELL=\"/bin/bash\"\n\n";
401: }
402: elsif ($mode eq 'configinstall') {
403: print '# LPML configuration file targets (configinstall).'."\n";
404: print '# Linux Packaging Markup Language,';
405: print ' by Scott Harrison 2001'."\n";
406: print '# This file was automatically generated on '.`date`;
407: print "\n".$invocation;
408: $lpml .= "SHELL=\"/bin/bash\"\n\n";
409: }
410: elsif ($mode eq 'build') {
411: $lpml = "# LPML build targets. Linux Packaging Markup Language,";
412: $lpml .= ' by Scott Harrison 2001'."\n";
413: $lpml .= '# This file was automatically generated on '.`date`;
414: $lpml .= "\n".$invocation;
415: $lpml .= "SHELL=\"/bin/sh\"\n\n";
416: }
417: else {
418: return '';
419: }
420: }
421: # --------------------------------------------------- Format targetroot section
422: sub format_targetroot {
423: my $text=&trim($parser->get_text('/targetroot'));
424: $text=$targetroot if $targetroot;
425: $parser->get_tag('/targetroot');
426: if ($mode eq 'html') {
427: return $targetroot="\n<br />TARGETROOT: $text";
428: }
429: elsif ($mode eq 'install' or $mode eq 'build' or
430: $mode eq 'configinstall') {
431: return '# TARGET INSTALL LOCATION is "'.$targetroot."\"\n";
432: }
433: else {
434: return '';
435: }
436: }
437: # --------------------------------------------------- Format sourceroot section
438: sub format_sourceroot {
439: my $text=&trim($parser->get_text('/sourceroot'));
440: $text=$sourceroot if $sourceroot;
441: $parser->get_tag('/sourceroot');
442: if ($mode eq 'html') {
443: return $sourceroot="\n<br />SOURCEROOT: $text";
444: }
445: elsif ($mode eq 'install' or $mode eq 'build' or
446: $mode eq 'configinstall') {
447: return '# SOURCE CODE LOCATION IS "'.$sourceroot."\"\n";;
448: }
449: else {
450: return '';
451: }
452: }
453: # --------------------------------------------------- Format categories section
454: sub format_categories {
455: my $text=&trim($parser->get_text('/categories'));
456: $parser->get_tag('/categories');
457: if ($mode eq 'html') {
458: return $categories="\n<br /> <br />".
459: "\n<a name='ownperms'>".
460: "\n<font size='+2'>File Type Ownership and Permissions".
461: " Descriptions</font>".
462: "\n<p>This table shows what permissions and ownership settings ".
463: "correspond to each category.</p>".
464: "\n<table border='1' cellpadding='5' width='60%'>\n".
465: "<tr>".
466: "<th align='left' bgcolor='#ffffff'>Icon</th>".
467: "<th align='left' bgcolor='#ffffff'>Category Name</th>".
468: "<th align='left' bgcolor='#ffffff'>Permissions ".
469: "($categorytype)</th>".
470: "</tr>".
471: "\n$text\n".
472: "</table>\n";
473: }
474: elsif ($mode eq 'text') {
475: return $categories="\n".
476: "\nFile Type Ownership and Permissions".
477: " Descriptions".
478: "\n$text".
479: "\n";
480: }
481: else {
482: return '';
483: }
484: }
485: # --------------------------------------------------- Format categories section
486: sub format_category {
487: my (@tokeninfo)=@_;
488: $category_att_name=$tokeninfo[2]->{'name'};
489: $category_att_type=$tokeninfo[2]->{'type'};
490: $abbreviation=''; $chmod='';$chown='';
491: $parser->get_text('/category');
492: $parser->get_tag('/category');
493: $fab{$category_att_name}=$abbreviation;
494: if ($mode eq 'html') {
495: if ($category_att_type eq $categorytype) {
496: return $category="<tr>".
497: "<td><img src='$abbreviation.gif' ".
498: "alt='${category_att_name}' /></td>\n".
499: "<td>${category_att_name}</td>\n".
500: "<td>$chmod $chown</td>\n".
501: "</tr>".
502: "\n";
503: # return $category="\n<br />CATEGORY $category_att_name ".
504: # "$category_att_type $chmod $chown";
505: }
506: }
507: else {
508: if ($category_att_type eq $categorytype) {
509: my ($user,$group)=split(/\:/,$chown);
510: $categoryhash{$category_att_name}='-o '.$user.' -g '.$group.
511: ' -m '.$chmod;
512: }
513: return '';
514: }
515: }
516: # --------------------------------------------------- Format categories section
517: sub format_abbreviation {
518: my @tokeninfo=@_;
519: $abbreviation='';
520: my $text=&trim($parser->get_text('/abbreviation'));
521: if ($text) {
522: $parser->get_tag('/abbreviation');
523: $abbreviation=$text;
524: }
525: return '';
526: }
527: # -------------------------------------------------------- Format chown section
528: sub format_chown {
529: my @tokeninfo=@_;
530: $chown='';
531: my $text=&trim($parser->get_text('/chown'));
532: if ($text) {
533: $parser->get_tag('/chown');
534: $chown=$text;
535: }
536: return '';
537: }
538: # -------------------------------------------------------- Format chmod section
539: sub format_chmod {
540: my @tokeninfo=@_;
541: $chmod='';
542: my $text=&trim($parser->get_text('/chmod'));
543: if ($text) {
544: $parser->get_tag('/chmod');
545: $chmod=$text;
546: }
547: return '';
548: }
549: # ---------------------------------------------------------- Format rpm section
550: sub format_rpm {
551: my $text=&trim($parser->get_text('/rpm'));
552: $parser->get_tag('/rpm');
553: if ($mode eq 'html') {
554: return $rpm=<<END;
555: <br /> <br />
556: <a name='package' />
557: <font size='+2'>Software Package Description</font>
558: <p>
559: <table bgcolor='#ffffff' border='0' cellpadding='10' cellspacing='0'>
560: <tr><td><pre>
561: $text
562: </pre></td></tr>
563: </table>
564: END
565: }
566: elsif ($mode eq 'text') {
567: return $rpm=<<END;
568: Software Package Description
569:
570: $text
571: END
572: }
573: else {
574: return '';
575: }
576: }
577: # --------------------------------------------------- Format rpmSummary section
578: sub format_rpmSummary {
579: my $text=&trim($parser->get_text('/rpmSummary'));
580: $parser->get_tag('/rpmSummary');
581: if ($mode eq 'html') {
582: return $rpmSummary="\nSummary : $text";
583: }
584: elsif ($mode eq 'text') {
585: return $rpmSummary="\nSummary : $text";
586: }
587: else {
588: return '';
589: }
590: }
591: # ------------------------------------------------------ Format rpmName section
592: sub format_rpmName {
593: my $text=&trim($parser->get_text('/rpmName'));
594: $parser->get_tag('/rpmName');
595: if ($mode eq 'html') {
596: return $rpmName="\nName : $text";
597: }
598: elsif ($mode eq 'text') {
599: return $rpmName="\nName : $text";
600: }
601: else {
602: return '';
603: }
604: }
605: # --------------------------------------------------- Format rpmVersion section
606: sub format_rpmVersion {
607: my $text=$parser->get_text('/rpmVersion');
608: $parser->get_tag('/rpmVersion');
609: if ($mode eq 'html') {
610: return $rpmVersion="\nVersion : $text";
611: }
612: elsif ($mode eq 'text') {
613: return $rpmVersion="\nVersion : $text";
614: }
615: else {
616: return '';
617: }
618: }
619: # --------------------------------------------------- Format rpmRelease section
620: sub format_rpmRelease {
621: my $text=$parser->get_text('/rpmRelease');
622: $parser->get_tag('/rpmRelease');
623: if ($mode eq 'html') {
624: return $rpmRelease="\nRelease : $text";
625: }
626: elsif ($mode eq 'text') {
627: return $rpmRelease="\nRelease : $text";
628: }
629: else {
630: return '';
631: }
632: }
633: # ---------------------------------------------------- Format rpmVendor section
634: sub format_rpmVendor {
635: my $text=$parser->get_text('/rpmVendor');
636: $parser->get_tag('/rpmVendor');
637: if ($mode eq 'html') {
638: return $rpmVendor="\nVendor : $text";
639: }
640: elsif ($mode eq 'text') {
641: return $rpmVendor="\nVendor : $text";
642: }
643: else {
644: return '';
645: }
646: }
647: # ------------------------------------------------- Format rpmBuildRoot section
648: sub format_rpmBuildRoot {
649: my $text=$parser->get_text('/rpmBuildRoot');
650: $parser->get_tag('/rpmBuildRoot');
651: if ($mode eq 'html') {
652: return $rpmBuildRoot="\nBuild Root : $text";
653: }
654: elsif ($mode eq 'text') {
655: return $rpmBuildRoot="\nBuild Root : $text";
656: }
657: else {
658: return '';
659: }
660: }
661: # ------------------------------------------------- Format rpmCopyright section
662: sub format_rpmCopyright {
663: my $text=$parser->get_text('/rpmCopyright');
664: $parser->get_tag('/rpmCopyright');
665: if ($mode eq 'html') {
666: return $rpmCopyright="\nLicense : $text";
667: }
668: elsif ($mode eq 'text') {
669: return $rpmCopyright="\nLicense : $text";
670: }
671: else {
672: return '';
673: }
674: }
675: # ----------------------------------------------------- Format rpmGroup section
676: sub format_rpmGroup {
677: my $text=$parser->get_text('/rpmGroup');
678: $parser->get_tag('/rpmGroup');
679: if ($mode eq 'html') {
680: return $rpmGroup="\nGroup : $text";
681: }
682: elsif ($mode eq 'text') {
683: return $rpmGroup="\nGroup : $text";
684: }
685: else {
686: return '';
687: }
688: }
689: # ---------------------------------------------------- Format rpmSource section
690: sub format_rpmSource {
691: my $text=$parser->get_text('/rpmSource');
692: $parser->get_tag('/rpmSource');
693: if ($mode eq 'html') {
694: return $rpmSource="\nSource : $text";
695: }
696: elsif ($mode eq 'text') {
697: return $rpmSource="\nSource : $text";
698: }
699: else {
700: return '';
701: }
702: }
703: # ----------------------------------------------- Format rpmAutoReqProv section
704: sub format_rpmAutoReqProv {
705: my $text=$parser->get_text('/rpmAutoReqProv');
706: $parser->get_tag('/rpmAutoReqProv');
707: if ($mode eq 'html') {
708: return $rpmAutoReqProv="\nAutoReqProv : $text";
709: }
710: if ($mode eq 'text') {
711: return $rpmAutoReqProv="\nAutoReqProv : $text";
712: }
713: else {
714: return '';
715: }
716: }
717: # ----------------------------------------------- Format rpmdescription section
718: sub format_rpmdescription {
719: my $text=$parser->get_text('/rpmdescription');
720: $parser->get_tag('/rpmdescription');
721: if ($mode eq 'html') {
722: $text=~s/\n//g;
723: $text=~s/\\n/\n/g;
724: return $rpmdescription="\nDescription : $text";
725: }
726: elsif ($mode eq 'text') {
727: $text=~s/\n//g;
728: $text=~s/\\n/\n/g;
729: return $rpmdescription="\nDescription : $text";
730: }
731: else {
732: return '';
733: }
734: }
735: # ------------------------------------------------------- Format rpmpre section
736: sub format_rpmpre {
737: my $text=$parser->get_text('/rpmpre');
738: $parser->get_tag('/rpmpre');
739: if ($mode eq 'html') {
740: # return $rpmpre="\n<br />RPMPRE $text";
741: return '';
742: }
743: else {
744: return '';
745: }
746: }
747: # -------------------------------------------------- Format directories section
748: sub format_directories {
749: my $text=$parser->get_text('/directories');
750: $parser->get_tag('/directories');
751: if ($mode eq 'html') {
752: $text=~s/\[\{\{\{\{\{DPATHLENGTH\}\}\}\}\}\]/$dpathlength/g;
753: return $directories="\n<br /> <br />".
754: "<a name='directories' />".
755: "<font size='+2'>Directory Structure</font>".
756: "\n<br /> <br />".
757: "<table border='1' cellpadding='3' cellspacing='0'>\n".
758: "<tr><th bgcolor='#ffffff'>Category</th>".
759: "<th bgcolor='#ffffff'>Status</th>\n".
760: "<th bgcolor='#ffffff'>Expected Permissions & Ownership</th>\n".
761: "<th bgcolor='#ffffff' colspan='$dpathlength'>Target Directory ".
762: "Path</th></tr>\n".
763: "\n$text\n</table><br />"."\n";
764: }
765: elsif ($mode eq 'text') {
766: return $directories="\nDirectory Structure\n$text\n".
767: "\n";
768: }
769: elsif ($mode eq 'install') {
770: return "\n".'directories:'."\n".$text;
771: }
772: else {
773: return '';
774: }
775: }
776: # ---------------------------------------------------- Format directory section
777: sub format_directory {
778: my (@tokeninfo)=@_;
779: $targetdir='';$categoryname='';$description='';
780: $parser->get_text('/directory');
781: $parser->get_tag('/directory');
782: if ($mode eq 'html') {
783: my @a;
784: @a=($targetdir=~/\//g);
785: my $d=scalar(@a)+1;
786: $dpathlength=$d if $d>$dpathlength;
787: my $thtml=$targetdir;
788: $thtml=~s/\//\<\/td\>\<td bgcolor='#ffffff'\>/g;
789: return $directory="\n<tr><td rowspan='2' bgcolor='#ffffff'>".
790: "$categoryname</td>".
791: "<td rowspan='2' bgcolor='#ffffff'><!-- POSTEVAL: --> </td>".
792: "<td rowspan='2' bgcolor='#ffffff'>$chmod<br />$chown</td>".
793: "<td bgcolor='#ffffff'>$thtml</td></tr>".
794: "<tr><td bgcolor='#ffffff' colspan='[{{{{{DPATHLENGTH}}}}}]'>".
795: "$description</td></tr>";
796: }
797: if ($mode eq 'text') {
798: return $directory="\nDIRECTORY $targetdir $categoryname ".
799: "$description";
800: }
801: elsif ($mode eq 'install') {
802: return "\t".'install '.$categoryhash{$categoryname}.' -d '.
803: $targetroot.'/'.$targetdir."\n";
804: }
805: else {
806: return '';
807: }
808: }
809: # ---------------------------------------------------- Format targetdir section
810: sub format_targetdir {
811: my @tokeninfo=@_;
812: $targetdir='';
813: my $text=&trim($parser->get_text('/targetdir'));
814: if ($text) {
815: $parser->get_tag('/targetdir');
816: $targetdir=$text;
817: }
818: return '';
819: }
820: # ------------------------------------------------- Format categoryname section
821: sub format_categoryname {
822: my @tokeninfo=@_;
823: $categoryname='';
824: my $text=&trim($parser->get_text('/categoryname'));
825: if ($text) {
826: $parser->get_tag('/categoryname');
827: $categoryname=$text;
828: }
829: return '';
830: }
831: # -------------------------------------------------- Format description section
832: sub format_description {
833: my @tokeninfo=@_;
834: $description='';
835: my $text=&htmlsafe(&trim($parser->get_text('/description')));
836: if ($text) {
837: $parser->get_tag('/description');
838: $description=$text;
839: }
840: return '';
841: }
842: # -------------------------------------------------------- Format files section
843: sub format_files {
844: my $text=$parser->get_text('/files');
845: $parser->get_tag('/files');
846: if ($mode eq 'html') {
847: return $directories="\n<br /> <br />".
848: "<a name='files' />".
849: "<font size='+2'>Files</font><br /> <br />".
850: "<p>All source and target locations are relative to the ".
851: "sourceroot and targetroot values at the beginning of this ".
852: "document.</p>".
853: "\n<table border='1' cellpadding='5'>".
854: "<tr><th>Status</th><th colspan='2'>Category</th>".
855: "<th>Name/Location</th>".
856: "<th>Description</th><th>Notes</th></tr>".
857: "$text</table>\n".
858: "\n";
859: }
860: elsif ($mode eq 'text') {
861: return $directories="\n".
862: "File and Directory Structure".
863: "\n$text\n".
864: "\n";
865: }
866: elsif ($mode eq 'install') {
867: return "\n".'files:'."\n".$text.
868: "\n".'links:'."\n".join('',@links);
869: }
870: elsif ($mode eq 'configinstall') {
871: return "\n".'configfiles: '.
872: join(' ',@configall).
873: "\n\n".$text.
874: "\n\nalwaysrun:\n\n";
875: }
876: elsif ($mode eq 'build') {
877: my $binfo;
878: my $tword;
879: my $command2;
880: my @deps;
881: foreach my $bi (@buildinfo) {
882: my ($target,$source,$command,$trigger,@deps)=split(/\;/,$bi);
883: $tword=''; $tword=' alwaysrun' if $trigger eq 'always run';
884: $command=~s/\/([^\/]*)$//;
885: $command2="cd $command; sh ./$1;\\";
886: my $depstring;
887: my $depstring2="\t\t\@echo '';\\\n";
888: my $olddep;
889: foreach my $dep (@deps) {
890: unless ($olddep) {
891: $olddep=$deps[$#deps];
892: }
893: $depstring.="\telif !(test -r $command/$dep);\\\n";
894: $depstring.="\t\tthen echo ".
895: "\"**** WARNING **** missing the file: ".
896: "$command/$dep\"$logcmd;\\\n";
897: $depstring.="\t\ttest -e $source || test -e $target || echo ".
898: "'**** ERROR **** neither source=$source nor target=".
899: "$target exist and they cannot be built'$logcmd;\\\n";
900: $depstring.="\t\tmake -f Makefile.build ${source}___DEPS;\\\n";
901: if ($olddep) {
902: $depstring2.="\t\tECODE=0;\\\n";
903: $depstring2.="\t\t! test -e $source && test -r $command/$olddep &&".
904: " { perl filecompare.pl -b2 $command/$olddep $target || ECODE=\$\$?; } && { [ \$\$ECODE != \"2\" ] || echo \"**** WARNING **** dependency $command/$olddep is newer than target file $target; SOMETHING MAY BE WRONG\"$logcmd; };\\\n";
905: }
906: $olddep=$dep;
907: }
908: $binfo.="$source: $tword\n".
909: "\t\@if !(echo \"\");\\\n\t\tthen echo ".
910: "\"**** WARNING **** Strange shell. ".
911: "Check your path settings.\"$logcmd;\\\n".
912: $depstring.
913: "\telse \\\n\t\t$command2\n\tfi\n\n";
914: $binfo.="${source}___DEPS:\n".$depstring2."\t\tECODE=0;\n\n";
915: }
916: return 'all: '.join(' ',@buildall)."\n\n".
917: $text.
918: $binfo."\n".
919: "alwaysrun:\n\n";
920: }
921: else {
922: return '';
923: }
924: }
925: # ---------------------------------------------------- Format fileglobs section
926: sub format_fileglobs {
927:
928: }
929: # -------------------------------------------------------- Format links section
930: # deprecated.. currently <link></link>'s are included in <files></files>
931: sub format_links {
932: my $text=$parser->get_text('/links');
933: $parser->get_tag('/links');
934: if ($mode eq 'html') {
935: return $links="\n<br />BEGIN LINKS\n$text\n<br />END LINKS\n";
936: }
937: elsif ($mode eq 'install') {
938: return "\n".'links:'."\n\t".$text;
939: }
940: else {
941: return '';
942: }
943: }
944: # --------------------------------------------------------- Format file section
945: sub format_file {
946: my @tokeninfo=@_;
947: $file=''; $source=''; $target=''; $categoryname=''; $description='';
948: $note=''; $build=''; $status=''; $dependencies='';
949: my $text=&trim($parser->get_text('/file'));
950: my $buildtest;
951: if ($source) {
952: $parser->get_tag('/file');
953: if ($mode eq 'html') {
954: return ($file="\n<!-- FILESORT:$target -->".
955: "<tr>".
956: "<td> </td><td><img src='$fab{$categoryname}.gif' ".
957: "alt='$categoryname icon' /></td>".
958: "<td>$categoryname</td>".
959: "<td>SOURCE: $source<br />TARGET: $target</td>".
960: "<td>$description</td>".
961: "<td>$note</td>".
962: "</tr>");
963: # return ($file="\n<br />BEGIN FILE\n".
964: # "$source $target $categoryname $description $note " .
965: # "$build $status $dependencies" .
966: # "\nEND FILE");
967: }
968: elsif ($mode eq 'install' && $categoryname ne 'conf') {
969: if ($build) {
970: my $bi=$sourceroot.'/'.$source.';'.$build.';'.
971: $dependencies;
972: my ($source2,$command,$trigger,@deps)=split(/\;/,$bi);
973: $tword=''; $tword=' alwaysrun' if $trigger eq 'always run';
974: $command=~s/\/([^\/]*)$//;
975: $command2="cd $command; sh ./$1;\\";
976: my $depstring;
977: foreach my $dep (@deps) {
978: $depstring.=<<END;
979: ECODE=0; DEP=''; \\
980: test -e $command/$dep || (echo '**** WARNING **** cannot evaluate status of dependency $command/$dep (for building ${sourceroot}/${source} with)'$logcmd); DEP="1"; \\
981: [ -n DEP ] && { perl filecompare.pl -b2 $command/$dep ${targetroot}/${target} || ECODE=\$\$?; } || DEP="1"; \\
982: case "\$\$ECODE" in \\
983: 2) echo "**** WARNING **** dependency $command/$dep is newer than target file ${targetroot}/${target}; you may want to run make build"$logcmd;; \\
984: esac; \\
985: END
986: }
987: chomp $depstring;
988: $buildtest=<<END;
989: \@if !(test -e "${sourceroot}/${source}") && !(test -e "${targetroot}/${target}"); then \\
990: echo "**** ERROR **** ${sourceroot}/${source} is missing and is also not present at target location ${targetroot}/${target}; you must run make build"$logcmd; exit; \\
991: END
992: $buildtest.=<<END if $depstring;
993: elif !(test -e "${sourceroot}/${source}"); then \\
994: $depstring
995: END
996: $buildtest.=<<END;
997: fi
998: END
999: }
1000: my $bflag='-b1';
1001: $bflag='-b3' if $dependencies or $buildlink;
1002: return <<END;
1003: $buildtest \@if !(test -e "${sourceroot}/${source}") && !(test -e "${targetroot}/${target}"); then \\
1004: echo "**** ERROR **** CVS source file does not exist: ${sourceroot}/${source} and neither does target: ${targetroot}/${target}"$logcmd; \\
1005: elif !(test -e "${sourceroot}/${source}"); then \\
1006: echo "**** WARNING **** CVS source file does not exist: ${sourceroot}/${source}"$logcmd; \\
1007: perl verifymodown.pl ${targetroot}/${target} "$categoryhash{$categoryname}"$logcmd; \\
1008: else \\
1009: ECODE=0; \\
1010: perl filecompare.pl $bflag ${sourceroot}/${source} ${targetroot}/${target} || ECODE=\$\$?; \\
1011: case "\$\$ECODE" in \\
1012: 1) echo "${targetroot}/${target} is unchanged";; \\
1013: 2) echo "**** WARNING **** target file ${targetroot}/${target} is newer than CVS source; saving current (old) target file to ${targetroot}/${target}.lpmlsave and then overwriting"$logcmd && install -o www -g www -m 0600 ${targetroot}/${target} ${targetroot}/${target}.lpmlsave && install $categoryhash{$categoryname} ${sourceroot}/${source} ${targetroot}/${target};; \\
1014: 0) echo "install $categoryhash{$categorname} ${sourceroot}/${source} ${targetroot}/${target}" && install $categoryhash{$categoryname} ${sourceroot}/${source} ${targetroot}/${target};; \\
1015: esac; \\
1016: perl verifymodown.pl ${targetroot}/${target} "$categoryhash{$categoryname}"$logcmd; \\
1017: fi
1018: END
1019: }
1020: elsif ($mode eq 'configinstall' && $categoryname eq 'conf') {
1021: push @configall,$targetroot.'/'.$target;
1022: return $targetroot.'/'.$target.': alwaysrun'."\n".
1023: "\t".'@echo -n ""; ECODE=0 && { perl filecompare.pl -b4 '.
1024: $sourceroot.'/'.$source.' '.$targetroot.'/'.$target.
1025: ' || ECODE=$$?; } && '.
1026: '{ [ $$ECODE != "2" ] || (install '.
1027: $categoryhash{$categoryname}.' '.
1028: $sourceroot.'/'.$source.' '.
1029: $targetroot.'/'.$target.'.lpmlnew'.
1030: ' && echo "**** NOTE: CONFIGURATION FILE CHANGE ****"'.
1031: $logcmd.' && echo "'.
1032: 'You likely need to compare contents of '.
1033: ''.$targetroot.'/'.$target.' with the new '.
1034: ''.$targetroot.'/'.$target.'.lpmlnew"'.
1035: "$logcmd); } && ".
1036: '{ [ $$ECODE != "3" ] || (install '.
1037: $categoryhash{$categoryname}.' '.
1038: $sourceroot.'/'.$source.' '.
1039: $targetroot.'/'.$target.''.
1040: ' && echo "**** WARNING: NEW CONFIGURATION FILE ADDED ****"'.
1041: $logcmd.' && echo "'.
1042: 'You likely need to review the contents of '.
1043: ''.$targetroot.'/'.$target.' to make sure its '.
1044: 'settings are compatible with your overall system"'.
1045: "$logcmd); } && ".
1046: '{ [ $$ECODE != "1" ] || ('.
1047: 'echo "**** ERROR ****"'.
1048: $logcmd.' && echo "'.
1049: 'Configuration source file does not exist '.
1050: ''.$sourceroot.'/'.$source.'"'.
1051: "$logcmd); } && perl verifymodown.pl ${targetroot}/${target} \"$categoryhash{$categoryname}\"$logcmd;\n\n";
1052: }
1053: elsif ($mode eq 'build' && $build) {
1054: push @buildall,$sourceroot.'/'.$source;
1055: push @buildinfo,$targetroot.'/'.$target.';'.$sourceroot.'/'.
1056: $source.';'.$build.';'.
1057: $dependencies;
1058: # return '# need to build '.$source.";
1059: }
1060: else {
1061: return '';
1062: }
1063: }
1064: return '';
1065: }
1066: # --------------------------------------------------------- Format link section
1067: sub format_link {
1068: my @tokeninfo=@_;
1069: $link=''; $linkto=''; $target=''; $categoryname=''; $description='';
1070: $note=''; $build=''; $status=''; $dependencies='';
1071: my $text=&trim($parser->get_text('/link'));
1072: if ($linkto) {
1073: $parser->get_tag('/link');
1074: if ($mode eq 'html') {
1075: return $link="\n<tr><td colspan='6'>BEGIN LINK\n".
1076: "$linkto $target $categoryname $description $note " .
1077: "$build $status $dependencies" .
1078: "\nEND LINK</td></tr>";
1079: }
1080: elsif ($mode eq 'install') {
1081: my @targets=map {s/^\s*//;s/\s$//;$_} split(/\;/,$target);
1082: foreach my $tgt (@targets) {
1083: push @links,"\t".'ln -fs /'.$linkto.' /'.$targetroot.$tgt.
1084: "\n";
1085: }
1086: return '';
1087: }
1088: else {
1089: return '';
1090: }
1091: }
1092: return '';
1093: }
1094: # ----------------------------------------------------- Format fileglob section
1095: sub format_fileglob {
1096: my @tokeninfo=@_;
1097: $fileglob=''; $glob=''; $sourcedir='';
1098: $targetdir=''; $categoryname=''; $description='';
1099: $note=''; $build=''; $status=''; $dependencies='';
1100: $filenames='';
1101: my $text=&trim($parser->get_text('/fileglob'));
1102: if ($sourcedir) {
1103: $parser->get_tag('/fileglob');
1104: if ($mode eq 'html') {
1105: return $fileglob="\n<tr><td colspan='6'>BEGIN FILEGLOB\n".
1106: "$glob sourcedir $targetdir $categoryname $description $note ".
1107: "$build $status $dependencies $filenames" .
1108: "\nEND FILEGLOB</td></tr>";
1109: }
1110: elsif ($mode eq 'install') {
1111: return "\t".'install '.
1112: $categoryhash{$categoryname}.' '.
1113: $sourceroot.'/'.$sourcedir.'[^C][^V][^S]'.$glob.' '.
1114: $targetroot.'/'.$targetdir.'.'."\n";
1115: }
1116: else {
1117: return '';
1118: }
1119: }
1120: return '';
1121: }
1122: # ---------------------------------------------------- Format sourcedir section
1123: sub format_sourcedir {
1124: my @tokeninfo=@_;
1125: $sourcedir='';
1126: my $text=&trim($parser->get_text('/sourcedir'));
1127: if ($text) {
1128: $parser->get_tag('/sourcedir');
1129: $sourcedir=$text;
1130: }
1131: return '';
1132: }
1133: # ------------------------------------------------------- Format target section
1134: sub format_target {
1135: my @tokeninfo=@_;
1136: $target='';
1137: my $text=&trim($parser->get_text('/target'));
1138: if ($text) {
1139: $parser->get_tag('/target');
1140: $target=$text;
1141: }
1142: return '';
1143: }
1144: # ------------------------------------------------------- Format source section
1145: sub format_source {
1146: my @tokeninfo=@_;
1147: $source='';
1148: my $text=&trim($parser->get_text('/source'));
1149: if ($text) {
1150: $parser->get_tag('/source');
1151: $source=$text;
1152: }
1153: return '';
1154: }
1155: # --------------------------------------------------------- Format note section
1156: sub format_note {
1157: my @tokeninfo=@_;
1158: $note='';
1159: # my $text=&trim($parser->get_text('/note'));
1160: my $aref;
1161: my $text;
1162: while ($aref=$parser->get_token()) {
1163: if ($aref->[0] eq 'E' && $aref->[1] eq 'note') {
1164: last;
1165: }
1166: elsif ($aref->[0] eq 'S') {
1167: $text.=$aref->[4];
1168: }
1169: elsif ($aref->[0] eq 'E') {
1170: $text.=$aref->[2];
1171: }
1172: else {
1173: $text.=$aref->[1];
1174: }
1175: }
1176: if ($text) {
1177: # $parser->get_tag('/note');
1178: $note=$text;
1179: }
1180: return '';
1181:
1182: }
1183: # -------------------------------------------------------- Format build section
1184: sub format_build {
1185: my @tokeninfo=@_;
1186: $build='';
1187: my $text=&trim($parser->get_text('/build'));
1188: if ($text) {
1189: $parser->get_tag('/build');
1190: $build=$sourceroot.'/'.$text.';'.$tokeninfo[2]{'trigger'};
1191: }
1192: return '';
1193: }
1194: # -------------------------------------------------------- Format build section
1195: sub format_buildlink {
1196: my @tokeninfo=@_;
1197: $buildlink='';
1198: my $text=&trim($parser->get_text('/buildlink'));
1199: if ($text) {
1200: $parser->get_tag('/buildlink');
1201: $buildlink=$sourceroot.'/'.$text;
1202: }
1203: return '';
1204: }
1205: # ------------------------------------------------------- Format status section
1206: sub format_status {
1207: my @tokeninfo=@_;
1208: $status='';
1209: my $text=&trim($parser->get_text('/status'));
1210: if ($text) {
1211: $parser->get_tag('/status');
1212: $status=$text;
1213: }
1214: return '';
1215: }
1216: # ------------------------------------------------- Format dependencies section
1217: sub format_dependencies {
1218: my @tokeninfo=@_;
1219: $dependencies='';
1220: my $text=&trim($parser->get_text('/dependencies'));
1221: if ($text) {
1222: $parser->get_tag('/dependencies');
1223: $dependencies=join(';',
1224: (map {s/^\s*//;s/\s$//;$_} split(/\;/,$text)));
1225: }
1226: return '';
1227: }
1228: # --------------------------------------------------------- Format glob section
1229: sub format_glob {
1230: my @tokeninfo=@_;
1231: $glob='';
1232: my $text=&trim($parser->get_text('/glob'));
1233: if ($text) {
1234: $parser->get_tag('/glob');
1235: $glob=$text;
1236: }
1237: return '';
1238: }
1239: # ---------------------------------------------------- Format filenames section
1240: sub format_filenames {
1241: my @tokeninfo=@_;
1242: my $text=&trim($parser->get_text('/filenames'));
1243: if ($text) {
1244: $parser->get_tag('/filenames');
1245: $filenames=$text;
1246: }
1247: return '';
1248: }
1249: # ------------------------------------------------------- Format linkto section
1250: sub format_linkto {
1251: my @tokeninfo=@_;
1252: my $text=&trim($parser->get_text('/linkto'));
1253: if ($text) {
1254: $parser->get_tag('/linkto');
1255: $linkto=$text;
1256: }
1257: return '';
1258: }
1259: # ------------------------------------- Render less-than and greater-than signs
1260: sub htmlsafe {
1261: my $text=@_[0];
1262: $text =~ s/</</g;
1263: $text =~ s/>/>/g;
1264: return $text;
1265: }
1266: # --------------------------------------- remove starting and ending whitespace
1267: sub trim {
1268: my ($s)=@_; $s=~s/^\s*//; $s=~s/\s*$//; return $s;
1269: }
1270:
1271: # ----------------------------------- POD (plain old documentation, CPAN style)
1272:
1273: =head1 NAME
1274:
1275: lpml_parse.pl - This is meant to parse files meeting the lpml document type.
1276: See lpml.dtd. LPML=Linux Packaging Markup Language.
1277:
1278: =head1 SYNOPSIS
1279:
1280: Usage is for lpml file to come in through standard input.
1281:
1282: =over 4
1283:
1284: =item *
1285:
1286: 1st argument is the mode of parsing.
1287:
1288: =item *
1289:
1290: 2nd argument is the category permissions to use (runtime or development)
1291:
1292: =item *
1293:
1294: 3rd argument is the distribution
1295: (default,redhat6.2,debian2.2,redhat7.1,etc).
1296:
1297: =item *
1298:
1299: 4th argument is to manually specify a sourceroot.
1300:
1301: =item *
1302:
1303: 5th argument is to manually specify a targetroot.
1304:
1305: =back
1306:
1307: Only the 1st argument is mandatory for the program to run.
1308:
1309: Example:
1310:
1311: cat ../../doc/loncapafiles.lpml |\\
1312: perl lpml_parse.pl html default /home/sherbert/loncapa /tmp/install
1313:
1314: =head1 DESCRIPTION
1315:
1316: I am using a multiple pass-through approach to parsing
1317: the lpml file. This saves memory and makes sure the server
1318: will never be overloaded.
1319:
1320: =head1 README
1321:
1322: I am using a multiple pass-through approach to parsing
1323: the lpml file. This saves memory and makes sure the server
1324: will never be overloaded.
1325:
1326: =head1 PREREQUISITES
1327:
1328: HTML::TokeParser
1329:
1330: =head1 COREQUISITES
1331:
1332: =head1 OSNAMES
1333:
1334: linux
1335:
1336: =head1 SCRIPT CATEGORIES
1337:
1338: Packaging/Administrative
1339:
1340: =cut
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>
![[BACK]](/icons/back.gif){kind=icon}
Return to lpml_parse.pl CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [LON-CAPA] / loncom / build