Adding API to map charset name to language. bug #26240 r=erik

intl/uconv/public/nsICharsetConverterManager.h

@@ -26,6 +26,7 @@
 #include "nsString.h"
 #include "nsError.h"
 #include "nsISupports.h"
+#include "nsIAtom.h"
 #include "nsIUnicodeEncoder.h"
 #include "nsIUnicodeDecoder.h"
 
@@ -39,33 +40,19 @@
 // XXX change to NS_CHARSETCONVERTERMANAGER_PID
 #define NS_CHARSETCONVERTERMANAGER_PROGID "charset-converter-manager"
 
-
+#define NS_REGISTRY_UCONV_BASE          "software/netscape/intl/uconv/"
-#define NS_REGISTRY_UCONV_BASE "software/netscape/intl/uconv/"
+// XXX change "xuconv" to "uconv" when the new enc&dec trees are in place
+#define NS_DATA_BUNDLE_REGISTRY_KEY     "software/netscape/intl/xuconv/data/"
+#define NS_TITLE_BUNDLE_REGISTRY_KEY    "software/netscape/intl/xuconv/titles/"
 
 #define NS_ERROR_UCONV_NOCONV \
   NS_ERROR_GENERATE_FAILURE(NS_ERROR_MODULE_UCONV, 0x01)
 
-#define SET_FOR_BROWSER(_val) (_val |= 0x00000001)
-#define SET_NOT_FOR_BROWSER(_val) (_val |= (!0x00000001))
-#define GET_FOR_BROWSER(_val) ((_val &= 0x00000001) != 0)
-
-#define SET_FOR_EDITOR(_val) (_val |= 0x00000001)
-#define SET_NOT_FOR_EDITOR(_val) (_val |= (!0x00000001))
-#define GET_FOR_EDITOR(_val) ((_val &= 0x00000001) != 0)
-
-#define SET_FOR_MAILNEWS(_val) (_val |= 0x00000002)
-#define SET_NOT_FOR_MAILNEWS(_val) (_val |= (!0x00000002))
-#define GET_FOR_MAILNEWS(_val) ((_val &= 0x00000002) != 0)
-
-#define SET_FOR_MAILNEWSEDITOR(_val) (_val |= 0x00000002)
-#define SET_NOT_FOR_MAILNEWSEDITOR(_val) (_val |= (!0x00000002))
-#define GET_FOR_MAILNEWSEDITOR(_val) ((_val &= 0x00000002) != 0)
-
 /**
  * Interface for a Manager of Charset Converters.
  * 
- * This Manager's data is a cacheing of Registry available stuff. But the 
+ * This Manager's data is a cache of the stuff available directly through 
- * access methods are also doing all the work to get it and provide it.
+ * Registry and Extensible String Bundles. Plus a set of convenient APIs.
  * 
  * Note: The term "Charset" used in the classes, interfaces and file names 
  * should be read as "Coded Character Set". I am saying "charset" only for 
@@ -76,6 +63,12 @@
  * A DECODER converts from a random encoding into Unicode.
  * An ENCODER converts from Unicode into a random encoding.
  * All our data structures and APIs are divided like that.
+ * However, when you have a charset data, you may have 3 cases:
+ * a) the data is charset-dependet, but it is common for encoders and decoders
+ * b) the data is different for the two of them, thus needing different APIs
+ * and different "aProp" identifying it.
+ * c) the data is relevant only for one: encoder or decoder; its nature making 
+ * the distinction.
  *
  * @created         15/Nov/1999
  * @author  Catalin Rotaru [CATA]
@@ -89,14 +82,18 @@ public:
       nsIUnicodeEncoder ** aResult) = 0;
   NS_IMETHOD GetUnicodeDecoder(const nsString * aSrc, 
       nsIUnicodeDecoder ** aResult) = 0;
+
   NS_IMETHOD GetDecoderList(nsString *** aResult, PRInt32 * aCount) = 0;
   NS_IMETHOD GetEncoderList(nsString *** aResult, PRInt32 * aCount) = 0;
-  NS_IMETHOD GetMIMEMailCharset(nsString * aCharset, nsString ** aResult) = 0;
+
-  NS_IMETHOD GetMIMEHeaderEncodingMethod(nsString * aCharset, 
-      nsString ** aResult) = 0;
   NS_IMETHOD GetCharsetData(nsString * aCharset, nsString * aProp, 
       nsString ** aResult) = 0;
   NS_IMETHOD GetCharsetTitle(nsString * aCharset, nsString ** aResult) = 0;
+  NS_IMETHOD GetCharsetLangGroup(nsString * aCharset, nsIAtom ** aResult) = 0;
+
+  NS_IMETHOD GetMIMEMailCharset(nsString * aCharset, nsString ** aResult) = 0;
+  NS_IMETHOD GetMIMEHeaderEncodingMethod(nsString * aCharset, 
+      nsString ** aResult) = 0;
 };
 
 #endif /* nsICharsetConverterManager_h___ */

intl/uconv/src/charsetData.properties

@@ -23,28 +23,56 @@
 ##    comparision in the code faster.
 
 ## Format of this file:
+##
 ## charset_name.notForBrowser = anything  -  specifies that this charset is 
 ## not to be used in the browser
 ##
+## charset_name.LangGroup = 
+##
 ## charset_name.MIMEHeaderEncodingMethod =
 ##
 ## charset_name.MIMEMailCharset =
 
-t.61-8bit.notForBrowser = true
+t.61-8bit.notForBrowser             = true
-utf-32le.notForBrowser = true
+utf-32le.notForBrowser              = true
-utf-32be.notForBrowser = true
+utf-32be.notForBrowser              = true
-utf-16le.notForBrowser = true
+utf-16le.notForBrowser              = true
-utf-16be.notForBrowser = true
+utf-16be.notForBrowser              = true
 x-imap4-modified-utf7.notForBrowser = true
-x-u-escaped.notForBrowser = true
+x-u-escaped.notForBrowser           = true
-us-ascii.notForBrowser = true
+hz-gb-2312.notForBrowser            = true
-x-obsoleted-euc-jp.notForBrowser = true
+x-gbk.notForBrowser                 = true
-x-obsoleted-iso-2022-jp.notForBrowser = true
+windows-936.notForBrowser           = true
-x-obsoleted-shift_jis.notForBrowser = true
+us-ascii.notForBrowser                  = true
-hz-gb-2312.notForBrowser = true
+x-obsoleted-euc-jp.notForBrowser        = true
-x-gbk.notForBrowser = true
+x-obsoleted-iso-2022-jp.notForBrowser   = true
-windows-936.notForBrowser = true
+x-obsoleted-shift_jis.notForBrowser     = true
 
-utf-8.MIMEHeaderEncodingMethod = B
 
-utf-8.MIMEMailCharset = utf-8
+shift_jis.LangGroup         = ja
+big5.LangGroup              = zh-TW
+euc-jp.LangGroup            = ja
+euc-kr.LangGroup            = ko
+gb2312.LangGroup            = zh-CN
+hz-gb-2312.LangGroup        = zh-CN
+iso-2022-jp.LangGroup       = ja
+iso-2022-kr.LangGroup       = ko
+iso-8859-1.LangGroup        = x-western
+iso-8859-2.LangGroup        = x-central-euro
+iso-8859-5.LangGroup        = x-cyrillic
+iso-8859-7.LangGroup        = el
+iso-8859-9.LangGroup        = tr
+koi8-r.LangGroup            = x-cyrillic
+utf-8.LangGroup             = x-unicode
+us-ascii.LangGroup          = x-western
+windows-1250.LangGroup      = x-central-euro
+windows-1251.LangGroup      = x-cyrillic
+x-euc-tw.LangGroup          = zh-TW
+x-mac-ce.LangGroup          = x-central-euro
+x-mac-cyrillic.LangGroup    = x-cyrillic
+x-mac-greek.LangGroup       = el
+x-mac-roman.LangGroup       = x-western
+x-mac-turkish.LangGroup     = tr
+
+utf-8.MIMEHeaderEncodingMethod  = B
+utf-8.MIMEMailCharset           = utf-8

intl/uconv/src/nsCharsetConverterManager.cpp

@@ -31,6 +31,7 @@
 #include "nsILocaleService.h"
 #include "nsUConvDll.h"
 #include "nsObjectArray.h"
+#include "prmem.h"
 
 // just for CIDs
 #include "nsIUnicodeDecodeHelper.h"
@@ -41,10 +42,6 @@ static NS_DEFINE_CID(kRegistryCID, NS_REGISTRY_CID);
 static NS_DEFINE_CID(kStringBundleServiceCID, NS_STRINGBUNDLESERVICE_CID);
 static NS_DEFINE_CID(kLocaleServiceCID, NS_LOCALESERVICE_CID); 
 
-// XXX change "xuconv" to "uconv" when the new enc&dec trees are in place
-#define DATA_BUNDLE_REGISTRY_KEY    "software/netscape/intl/xuconv/data/"
-#define TITLE_BUNDLE_REGISTRY_KEY   "software/netscape/intl/xuconv/titles/"
-
 //----------------------------------------------------------------------------
 // Class nsConverterInfo [declaration]
 
@@ -99,6 +96,9 @@ private:
   nsresult GetBundleValue(nsIStringBundle * aBundle, nsString * aName, 
       nsString * aProp, nsString ** aResult);
 
+  nsresult GetBundleValue(nsIStringBundle * aBundle, nsString * aName, 
+      nsString * aProp, nsIAtom ** aResult);
+
 public:
 
   nsCharsetConverterManager();
@@ -113,14 +113,18 @@ public:
       nsIUnicodeEncoder ** aResult);
   NS_IMETHOD GetUnicodeDecoder(const nsString * aSrc, 
       nsIUnicodeDecoder ** aResult);
+
   NS_IMETHOD GetDecoderList(nsString *** aResult, PRInt32 * aCount);
   NS_IMETHOD GetEncoderList(nsString *** aResult, PRInt32 * aCount);
-  NS_IMETHOD GetMIMEMailCharset(nsString * aCharset, nsString ** aResult);
+
-  NS_IMETHOD GetMIMEHeaderEncodingMethod(nsString * aCharset, nsString ** 
-      aResult);
   NS_IMETHOD GetCharsetData(nsString * aCharset, nsString * aProp, 
       nsString ** aResult);
   NS_IMETHOD GetCharsetTitle(nsString * aCharset, nsString ** aResult);
+  NS_IMETHOD GetCharsetLangGroup(nsString * aCharset, nsIAtom ** aResult);
+
+  NS_IMETHOD GetMIMEMailCharset(nsString * aCharset, nsString ** aResult);
+  NS_IMETHOD GetMIMEHeaderEncodingMethod(nsString * aCharset, nsString ** 
+      aResult);
 };
 
 //----------------------------------------------------------------------------
@@ -204,8 +208,8 @@ nsresult nsCharsetConverterManager::RegisterConverterManagerData()
     if (NS_FAILED(res)) return res;
   }
 
-  RegisterConverterTitles(registry, TITLE_BUNDLE_REGISTRY_KEY);
+  RegisterConverterTitles(registry, NS_TITLE_BUNDLE_REGISTRY_KEY);
-  RegisterConverterData(registry, DATA_BUNDLE_REGISTRY_KEY);
+  RegisterConverterData(registry, NS_DATA_BUNDLE_REGISTRY_KEY);
 
   return NS_OK;
 }
@@ -404,18 +408,36 @@ nsresult nsCharsetConverterManager::GetBundleValue(nsIStringBundle * aBundle,
 {
   nsresult res = NS_OK;
 
-  nsAutoString key(aName->GetUnicode());
+  nsAutoString key(*aName);
-  key.Append(*aProp);
+  if (aProp != NULL) key.Append(*aProp); // yes, this parameter may be NULL
 
   PRUnichar * value = NULL;
   res = aBundle->GetStringFromName(key.GetUnicode(), &value);
   if (NS_FAILED(res)) return res;
 
   *aResult = new nsString(value);
-  delete value;
+  PR_Free(value);
   return res;
 }
 
+nsresult nsCharsetConverterManager::GetBundleValue(nsIStringBundle * aBundle, 
+                                                   nsString * aName, 
+                                                   nsString * aProp, 
+                                                   nsIAtom ** aResult)
+{
+  nsresult res = NS_OK;
+
+  nsAutoString key(*aName);
+  if (aProp != NULL) key.Append(*aProp); // yes, this parameter may be NULL
+
+  PRUnichar * value = NULL;
+  res = aBundle->GetStringFromName(key.GetUnicode(), &value);
+  if (NS_FAILED(res)) return res;
+
+  *aResult =  NS_NewAtom(value);
+  PR_Free(value);
+  return res;
+}
 
 //----------------------------------------------------------------------------
 // Interface nsICharsetConverterManager [implementation]
@@ -478,10 +500,14 @@ NS_IMETHODIMP nsCharsetConverterManager::GetCharsetData(nsString * aCharset,
                                                         nsString * aProp, 
                                                         nsString ** aResult)
 {
-  nsresult res = NS_OK;;
+  if (aCharset == NULL) return NS_ERROR_NULL_POINTER;
+  if (aResult == NULL) return NS_ERROR_NULL_POINTER;
+  *aResult = NULL;
+
+  nsresult res = NS_OK;
 
   if (mDataBundle == NULL) {
-    res = LoadExtensibleBundle(DATA_BUNDLE_REGISTRY_KEY, &mDataBundle);
+    res = LoadExtensibleBundle(NS_DATA_BUNDLE_REGISTRY_KEY, &mDataBundle);
     if (NS_FAILED(res)) return res;
   }
 
@@ -492,11 +518,15 @@ NS_IMETHODIMP nsCharsetConverterManager::GetCharsetData(nsString * aCharset,
 NS_IMETHODIMP nsCharsetConverterManager::GetCharsetTitle(nsString * aCharset, 
                                                          nsString ** aResult)
 {
-  nsresult res = NS_OK;;
+  if (aCharset == NULL) return NS_ERROR_NULL_POINTER;
+  if (aResult == NULL) return NS_ERROR_NULL_POINTER;
+  *aResult = NULL;
+
+  nsresult res = NS_OK;
   nsAutoString prop(".title");
 
   if (mTitleBundle == NULL) {
-    res = LoadExtensibleBundle(TITLE_BUNDLE_REGISTRY_KEY, &mTitleBundle);
+    res = LoadExtensibleBundle(NS_TITLE_BUNDLE_REGISTRY_KEY, &mTitleBundle);
     if (NS_FAILED(res)) return res;
   }
 
@@ -504,6 +534,26 @@ NS_IMETHODIMP nsCharsetConverterManager::GetCharsetTitle(nsString * aCharset,
   return res;
 }
 
+NS_IMETHODIMP nsCharsetConverterManager::GetCharsetLangGroup(
+                                         nsString * aCharset, 
+                                         nsIAtom ** aResult)
+{
+  if (aCharset == NULL) return NS_ERROR_NULL_POINTER;
+  if (aResult == NULL) return NS_ERROR_NULL_POINTER;
+  *aResult = NULL;
+
+  nsresult res = NS_OK;;
+  nsAutoString prop(".LangGroup");
+
+  if (mDataBundle == NULL) {
+    res = LoadExtensibleBundle(NS_DATA_BUNDLE_REGISTRY_KEY, &mDataBundle);
+    if (NS_FAILED(res)) return res;
+  }
+
+  res = GetBundleValue(mDataBundle, aCharset, &prop, aResult);
+  return res;
+}
+
 NS_IMETHODIMP nsCharsetConverterManager::GetMIMEMailCharset(
                                          nsString * aCharset, 
                                          nsString ** aResult)

intl/uconv/src/nsCharsetMenu.cpp

@@ -116,6 +116,9 @@ private:
   nsresult AddItemToContainer(nsIRDFService * aRDFServ, 
       nsICharsetConverterManager * aCCMan, nsIRDFContainer * aContainer, 
       nsMenuItem * aItem);
+  nsresult AddFromStringToMenu(char * aCharsetList, 
+      nsIRDFService * aRDFServ, nsICharsetConverterManager * aCCMan, 
+      nsObjectArray * aObjectArray, nsIRDFContainer * aContainer);
   nsresult AddFromPrefsToMenu(nsIPref * aPref, nsIRDFService * aRDFServ, 
       nsICharsetConverterManager * aCCMan, nsObjectArray * aObjectArray, 
       nsIRDFContainer * aContainer, char * aKey);
@@ -532,6 +535,31 @@ nsresult nsCharsetMenu::AddItemToContainer(nsIRDFService * aRDFServ,
   return res;
 }
 
+nsresult nsCharsetMenu::AddFromStringToMenu(char * aCharsetList,
+                                            nsIRDFService * aRDFServ, 
+                                            nsICharsetConverterManager * aCCMan, 
+                                            nsObjectArray * aObjectArray, 
+                                            nsIRDFContainer * aContainer)
+{
+  char * p = aCharsetList;
+  char * q = p;
+  while (*p != 0) {
+    for (; (*q != ',') && (*q != ' ') && (*q != 0); q++);
+    char temp = *q;
+    *q = 0;
+
+    nsAutoString str(p);
+    nsresult res = AddItemToMenu(aRDFServ, aCCMan, aObjectArray, aContainer, &str);
+    NS_ASSERTION(NS_SUCCEEDED(res), "failed to add item to menu");
+
+    *q = temp;
+    for (; (*q == ',') || (*q == ' '); q++);
+    p=q;
+  }
+
+  return NS_OK;
+}
+
 nsresult nsCharsetMenu::AddFromPrefsToMenu(nsIPref * aPref, 
                                            nsIRDFService * aRDFServ, 
                                            nsICharsetConverterManager * aCCMan, 
@@ -546,21 +574,7 @@ nsresult nsCharsetMenu::AddFromPrefsToMenu(nsIPref * aPref,
   if (NS_FAILED(res)) return res;
 
   if (value != NULL) {
-    char * p = value;
+    res = AddFromStringToMenu(value, aRDFServ, aCCMan, aObjectArray, aContainer);
-    char * q = p;
-    while (*p != 0) {
-      for (; (*q != ',') && (*q != ' ') && (*q != 0); q++);
-      char temp = *q;
-      *q = 0;
-
-      nsAutoString str(p);
-      AddItemToMenu(aRDFServ, aCCMan, aObjectArray, aContainer, &str);
-
-      *q = temp;
-      for (; (*q == ',') || (*q == ' '); q++);
-      p=q;
-    }
-
     nsAllocator::Free(value);
   }