Many changes to make output test really useful. Fix Windows makefile. Add some sample test input files

git-svn-id: svn://10.0.0.236/trunk@49318 18797224-902f-48f8-a5cc-f745e15eee43

mozilla/htmlparser/tests/outsinks/Convert.cpp

@@ -20,12 +20,15 @@
  * Contributor(s): Akkana Peck.
  */
 
+#include <ctype.h>      // for isdigit()
+
 #include "nsParserCIID.h"
 #include "nsIParser.h"
 #include "nsHTMLContentSinkStream.h"
 #include "nsHTMLToTXTSinkStream.h"
 #include "nsIComponentManager.h"
 #include "CNavDTD.h"
+#include "nsXIFDTD.h"
 
 extern "C" void NS_SetupRegistry();
 
@@ -44,27 +47,26 @@ static NS_DEFINE_IID(kParserCID, NS_PARSER_IID);
 // Interface IID's
 static NS_DEFINE_IID(kIParserIID, NS_IPARSER_IID);
 
+nsresult
+Compare(nsString& str, nsString& filename)
+{
+  printf("Sorry, don't know how to compare yet\n");
+  char* charstar = str.ToNewUTF8String();
+  printf("Output string is: %s\n-------------------- \n", charstar);
+  delete[] charstar;
+  return NS_ERROR_NOT_IMPLEMENTED;
+}
+
 //----------------------------------------------------------------------
 // Convert html on stdin to either plaintext or (if toHTML) html
 //----------------------------------------------------------------------
-nsresult HTML2text(int toHTML)
+nsresult
+HTML2text(nsString& inString, nsString& inType, nsString& outType, int wrapCol, nsString& compareAgainst)
 {
   nsresult rv = NS_OK;
 
-  nsString inString;
   nsString outString;
 
-  // Read in the string from the file: very inefficient, but who cares?
-  char c;
-  while ((c = getchar()) != EOF)
-    inString += c;
-
-#if 0
-  printf("Input string is: %s\n-------------------- \n",
-         inString.ToNewCString());
-  printf("------------------------------------\n");
-#endif
-
   // Create a parser
   nsIParser* parser;
    rv = nsComponentManager::CreateInstance(kParserCID, nsnull,
@@ -78,11 +80,11 @@ nsresult HTML2text(int toHTML)
   nsIHTMLContentSink* sink = nsnull;
 
   // Create the appropriate output sink
-  if (toHTML)
+  if (outType == "text/html")
     rv = NS_New_HTML_ContentSinkStream(&sink, &outString, 0);
 
   else  // default to plaintext
-    rv = NS_New_HTMLToTXT_SinkStream(&sink, &outString, 72, 2);
+    rv = NS_New_HTMLToTXT_SinkStream(&sink, &outString, wrapCol, 2);
 
   if (NS_FAILED(rv))
   {
@@ -92,7 +94,10 @@ nsresult HTML2text(int toHTML)
 
   parser->SetContentSink(sink);
   nsIDTD* dtd = nsnull;
-  rv = NS_NewNavHTMLDTD(&dtd);
+  if (inType == "text/xif")
+    rv = NS_NewXIFDTD(&dtd);
+  else
+    rv = NS_NewNavHTMLDTD(&dtd);
   if (NS_FAILED(rv))
   {
     printf("Couldn't create new HTML DTD: 0x%x\n", rv);
@@ -101,32 +106,138 @@ nsresult HTML2text(int toHTML)
 
   parser->RegisterDTD(dtd);
 
-  rv = parser->Parse(inString, 0, "text/html", PR_FALSE, PR_TRUE);
+  char* inTypeStr = inType.ToNewCString();
+  rv = parser->Parse(inString, 0, inTypeStr, PR_FALSE, PR_TRUE);
+  delete[] inTypeStr;
   if (NS_FAILED(rv))
   {
     printf("Parse() failed! 0x%x\n", rv);
     return rv;
   }
 
-  printf("Output string is: %s\n-------------------- \n",
-         outString.ToNewCString());
-
   NS_IF_RELEASE(dtd);
   NS_IF_RELEASE(sink);
   NS_RELEASE(parser);
 
-  return rv;
+  if (compareAgainst.Length() > 0)
+    return Compare(outString, compareAgainst);
+
+  char* charstar = outString.ToNewUTF8String();
+  printf("Output string is:\n--------------------\n%s\n--------------------\n",
+         charstar);
+  delete[] charstar;
+
+  return NS_OK;
 }
 
 //----------------------------------------------------------------------
 
 int main(int argc, char** argv)
 {
+  nsString inType ("text/html");
+  nsString outType ("text/plain");
+  int wrapCol = 72;
+  nsString compareAgainst;
+
+
+  // Skip over progname arg:
+  const char* progname = argv[0];
+  --argc; ++argv;
+
+  // Process flags
+  while (argc > 0 && argv[0][0] == '-')
+  {
+    switch (argv[0][1])
+    {
+      case 'h':
+        printf("\
+Usage: %s [-i intype] [-o outtype] [-w wrapcol] [-c comparison_file] infile\n\
+\tIn/out types are mime types (e.g. text/html)\n\
+\tcomparison_file is a file against which to compare the output\n\
+\t  (not yet implemented\n\
+\tDefaults are -i text/html -o text/plain -w 72 [stdin]\n",
+               progname);
+        exit(0);
+
+        case 'i':
+        if (argv[0][2] != '\0')
+          inType = argv[0]+2;
+        else {
+          inType = argv[1];
+          --argc;
+          ++argv;
+        }
+        break;
+
+      case 'o':
+        if (argv[0][2] != '\0')
+          outType = argv[0]+2;
+        else {
+          outType = argv[1];
+          --argc;
+          ++argv;
+        }
+        break;
+
+      case 'w':
+        if (isdigit(argv[0][2]))
+          wrapCol = atoi(argv[0]+2);
+        else {
+          wrapCol = atoi(argv[1]);
+          --argc;
+          ++argv;
+        }
+        break;
+
+      case 'c':
+        if (argv[0][2] != '\0')
+          compareAgainst = argv[0]+2;
+        else {
+          compareAgainst = argv[1];
+          --argc;
+          ++argv;
+        }
+        break;
+    }
+    ++argv;
+    --argc;
+  }
+
+  FILE* file = 0;
+  if (argc > 0)         // read from a file
+  {
+    // Open the file in a Unix-centric way,
+    // until I find out how to use nsFileSpec:
+    file = fopen(argv[0], "r");
+    if (!file)
+    {
+      fprintf(stderr, "Can't open file %s", argv[0]);
+      perror(" ");
+      exit(1);
+    }
+  }
+  else file = stdin;
+
   nsComponentManager::AutoRegister(nsIComponentManager::NS_Startup, 0);
   NS_SetupRegistry();
 
-  if (argc > 1 && argv[1][0] == '-' && argv[1][1] == 'h')
-    HTML2text(1);
-  else
-    HTML2text(0);
+  // Read in the string: very inefficient, but who cares?
+  nsString inString;
+  char c;
+  while ((c = getc(file)) != EOF)
+    inString += c;
+
+  if (file != stdin)
+    fclose(file);
+
+#if 0
+  printf("Input string is: %s\n-------------------- \n",
+         inString.ToNewCString());
+  printf("------------------------------------\n");
+#endif
+
+  printf("inType = '%s', outType = '%s'\n", inType.ToNewCString(), outType.ToNewCString());
+  HTML2text(inString, inType, outType, wrapCol, compareAgainst);
+
+  return 0;
 }