From 8e27b343e0b8a7e8710bdc7748ddac3a02b0a45c Mon Sep 17 00:00:00 2001
From: ulatekh <ulatekh@yahoo.com>
Date: Sun, 29 Jul 2018 09:07:42 -0700
Subject: [PATCH] Now pdftohtml takes a "-pg" parameter, with a list of page
 ranges. This is more flexible and powerful than -f and -l, which only allows
 one page range.

---
 utils/pdftohtml.cc | 109 +++++++++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 106 insertions(+), 3 deletions(-)

diff --git a/utils/pdftohtml.cc b/utils/pdftohtml.cc
index 04aeb1bc..32a2d265 100644
--- a/utils/pdftohtml.cc
+++ b/utils/pdftohtml.cc
@@ -66,9 +66,13 @@
 #include "DateInfo.h"
 #include "goo/gfile.h"
 #include "Win32Console.h"
+#include <boost/algorithm/string.hpp>
+#include <boost/icl/interval_set.hpp>
 
 static int firstPage = 1;
 static int lastPage = 0;
+static GooString pageRange;
+static boost::icl::interval_set<int> setPageRanges;
 static GBool rawOrder = gTrue;
 GBool printCommands = gTrue;
 static GBool printHelp = gFalse;
@@ -102,6 +106,8 @@ static const ArgDesc argDesc[] = {
    "first page to convert"},
   {"-l",      argInt,      &lastPage,      0,
    "last page to convert"},
+  {"-pg",     argGooString,&pageRange,     0,
+   "range of pages to convert (e.g. \"1-2,5-6,19-\")"},
   /*{"-raw",    argFlag,     &rawOrder,      0,
     "keep strings in content stream order"},*/
   {"-q",      argFlag,     &errQuiet,      0,
@@ -291,7 +297,7 @@ int main(int argc, char *argv[]) {
     }
     delete tmp;
   } else if (fileName->cmp("fd://0") == 0) {
-      error(errCommandLine, -1, "You have to provide an output filename when reading form stdin.");
+      error(errCommandLine, -1, "You have to provide an output filename when reading from stdin.");
       goto error;
   } else {
     p = fileName->getCString() + fileName->getLength() - 4;
@@ -337,6 +343,98 @@ int main(int argc, char *argv[]) {
     goto error;
   }
 
+  // get page ranges
+  if (pageRange.getLength() == 0)
+  {
+    setPageRanges.add(boost::icl::interval<int>::type(
+      firstPage, lastPage + 1));
+  }
+  else
+  {
+    // Prepare to build a set of page-ranges from a
+    // comma-delimited string.
+    std::vector<std::string> vecPageRanges;
+    char const *pszPageRanges = pageRange.getCString();
+      int iPageCount = doc->getNumPages();
+    boost::split(vecPageRanges, pszPageRanges,
+      boost::is_any_of(","));
+    for(std::string const &rstrPageRange : vecPageRanges)
+    {
+      // A range is two numbers, separated by a dash.
+      // If there's no first number, the beginning of
+      // the range is the first page in the document.
+      // If there's no second number, the end of the
+      // range is the last page in the document.
+      std::vector<std::string> vecPageRange;
+      char const *pszPageRange = rstrPageRange.c_str();
+      boost::split(vecPageRange, pszPageRange,
+        boost::is_any_of("-"));
+      if (vecPageRange.size() < 1 || vecPageRange.size() > 2)
+      {
+        error(errCommandLine, -1,
+          "Invalid page range given: \"{0:s}\"",
+          pszPageRange);
+        goto error;
+      }
+
+      // Parse the first number in the range.
+      // If it's empty, it's the first page in the document.
+      int iFirstPage = 0;
+      {
+        std::string const &rstrFirstPage = vecPageRange[0];
+        char const *pszFirstPage = rstrFirstPage.c_str();
+        char const *pszFirstPageEnd
+          = pszFirstPage + rstrFirstPage.size();
+        if (pszFirstPage == pszFirstPageEnd)
+          iFirstPage = 1;
+        else
+        {
+          char /* const */ *pszValueEnd = nullptr;
+          iFirstPage = ::strtol(pszFirstPage, &pszValueEnd, 10);
+          if (pszValueEnd != pszFirstPageEnd)
+          {
+            error(errCommandLine, -1,
+              "Page-range value \"{0:s}\" is not a valid number",
+              pszFirstPage);
+            goto error;
+          }
+        }
+      }
+
+      // Parse the last number in the range.
+      // If it's missing, it's a one-page range.
+      // If it's empty, it's the last page in the document.
+      int iLastPage = 0;
+      if (vecPageRange.size() < 2)
+        iLastPage = iFirstPage;
+      else
+      {
+        std::string const &rstrLastPage = vecPageRange[1];
+        char const *pszLastPage = rstrLastPage.c_str();
+        char const *pszLastPageEnd
+          = pszLastPage + rstrLastPage.size();
+        if (pszLastPage == pszLastPageEnd)
+          iLastPage = iPageCount;
+        else
+        {
+          char /* const */ *pszValueEnd = nullptr;
+          iLastPage = ::strtol(pszLastPage, &pszValueEnd, 10);
+          if (pszValueEnd != pszLastPageEnd)
+          {
+            error(errCommandLine, -1,
+              "Page-range value \"{0:s}\" is not a valid number",
+              pszLastPage);
+            goto error;
+          }
+        }
+      }
+
+      // Accumulate this range of pages.
+      setPageRanges.add(boost::icl::interval<int>::type(
+        iFirstPage, iLastPage + 1));
+    }
+  }
+
   info = doc->getDocInfo();
   if (info.isDict()) {
     docTitle = getInfoString(info.getDict(), "Title");
@@ -390,8 +488,13 @@ int main(int argc, char *argv[]) {
 
   if (htmlOut->isOk())
   {
-    doc->displayPages(htmlOut, firstPage, lastPage, 72 * scale, 72 * scale, 0,
-		      gTrue, gFalse, gFalse);
+    for (auto const &rPageRange : setPageRanges)
+    {
+      int iFirstPage = rPageRange.lower();
+      int iLastPage = rPageRange.upper() - 1;
+      doc->displayPages(htmlOut, iFirstPage, iLastPage, 72 * scale, 72 * scale, 0,
+		        gTrue, gFalse, gFalse);
+    }
     htmlOut->dumpDocOutline(doc);
   }
   
-- 
2.14.4