added external filters and pdf handling

2005-02-01 17:20:06 +00:00
parent cc512e2ec0
commit d0aaf92220
14 changed files with 373 additions and 82 deletions
@@ -0,0 +1,84 @@
+#!/bin/sh
+# @(#$Id: rclpdf,v 1.1 2005-02-01 17:20:05 dockes Exp $  (C) 2004 J.F.Dockes
+# This is copied almost verbatim from Estraier:
+#================================================================
+# Estraier: a personal full-text search system
+# Copyright (C) 2003-2004 Mikio Hirabayashi
+#================================================================
+#================================================================
+# rclpdf
+# Strip a file of PDF and extract its text as HTML.
+#================================================================
+
+
+# set variables
+LANG=C ; export LANG
+LC_ALL=C ; export LC_ALL
+progname="rclpdf"
+
+# show help message
+if test $# -ne 1 -o "$1" = "--help" 
+then
+  printf 'Strip a file of PDF and extract its text as HTML.\n'
+  printf 'Usage: %s [infile]\n' "$progname"
+  exit 1
+fi
+
+infile="$1"
+
+# check the input file existence
+if test ! -f "$infile"
+then
+  printf '%s: %s: no such file\n' "$progname" "$infile"
+  exit 1
+fi
+
+# output the result
+pdftotext -raw -htmlmeta -enc UTF-8 -eol unix -q "$infile" - |
+iconv -f UTF-8 -t UTF-8 -c -s |
+awk '
+BEGIN {
+  esc = 0
+  mul = 1
+  emp = 0
+}
+{
+  if(esc < 1 && $0 ~ /^<title>/ && $0 ~ /title>$/){
+    printf("<meta http-equiv=\"Content-Type\" content=\"text/html; charset=UTF-8\">\n")
+    gsub(/<[^>]*>/, "", $0)
+    gsub(/&/, "\\&amp;", $0)
+    gsub(/</, "\\&lt;", $0)
+    gsub(/>/, "\\&gt;", $0)
+    printf("<title>%s</title>\n", $0)
+  } else if($0 == "<pre>"){
+    esc++
+    printf("<p>")
+    mul = 1
+  } else if($0 == "</pre>"){
+    esc--
+    printf("</p>\n")
+  } else if($0 ~ /-$/){
+    sub(/-$/, "", $0)
+    printf("%s", $0);
+  } else if($0 == "\f"){
+    printf("</p>\n<hr>\n<p>")
+  } else {
+    if(esc > 0){
+      gsub(/&/, "\\&amp;", $0)
+      gsub(/</, "\\&lt;", $0)
+      gsub(/>/, "\\&gt;", $0)
+      gsub(/^ */, "", $0)
+      gsub(/ *$/, "", $0)
+    }
+    print $0
+  }
+}
+' 
+# Suppressed code 2 lines above (at the last print $0), which seemed to
+# deal with multibyte character being cut by a newline ? It caused problems
+# (sometimes concatenated last word of a line with first of next, and I
+# didn't really understand its use as iconv -c is supposed to fix the
+# encoding anyway
+
+# exit normally
+exit 0