From da5404f61273043f40b61ac3c1de9cc03bf441f5 Mon Sep 17 00:00:00 2001
From: Michael Jumper <mike.jumper@guac-dev.org>
Date: Wed, 2 Apr 2014 14:08:41 -0700
Subject: [PATCH] GUAC-584: Add support for CP1252 and ISO-8859-1.

---
 src/common/guac_iconv.c   | 96 +++++++++++++++++++++++++++++++++++++++
 src/common/guac_iconv.h   | 20 ++++++++
 tests/common/guac_iconv.c | 36 +++++++++++++++
 3 files changed, 152 insertions(+)

diff --git a/src/common/guac_iconv.c b/src/common/guac_iconv.c
index 0dcab980..00f07661 100644
--- a/src/common/guac_iconv.c
+++ b/src/common/guac_iconv.c
@@ -26,6 +26,44 @@
 #include <guacamole/unicode.h>
 #include <stdint.h>
 
+/**
+ * Lookup table for Unicode code points, indexed by CP-1252 codepoint.
+ */
+const static int __GUAC_RDP_CP1252_CODEPOINT[32] = {
+    0x20AC, /* 0x80 */
+    0xFFFD, /* 0x81 */
+    0x201A, /* 0x82 */
+    0x0192, /* 0x83 */
+    0x201E, /* 0x84 */
+    0x2026, /* 0x85 */
+    0x2020, /* 0x86 */
+    0x2021, /* 0x87 */
+    0x02C6, /* 0x88 */
+    0x2030, /* 0x89 */
+    0x0160, /* 0x8A */
+    0x2039, /* 0x8B */
+    0x0152, /* 0x8C */
+    0xFFFD, /* 0x8D */
+    0x017D, /* 0x8E */
+    0xFFFD, /* 0x8F */
+    0xFFFD, /* 0x90 */
+    0x2018, /* 0x91 */
+    0x2019, /* 0x92 */
+    0x201C, /* 0x93 */
+    0x201D, /* 0x94 */
+    0x2022, /* 0x95 */
+    0x2013, /* 0x96 */
+    0x2014, /* 0x97 */
+    0x02DC, /* 0x98 */
+    0x2122, /* 0x99 */
+    0x0161, /* 0x9A */
+    0x203A, /* 0x9B */
+    0x0153, /* 0x9C */
+    0xFFFD, /* 0x9D */
+    0x017E, /* 0x9E */
+    0x0178, /* 0x9F */
+};
+
 int guac_iconv(guac_iconv_read* reader, char** input, int in_remaining,
                guac_iconv_write* writer, char** output, int out_remaining) {
 
@@ -81,6 +119,28 @@ int GUAC_READ_UTF16(char** input, int remaining) {
 
 }
 
+int GUAC_READ_CP1252(char** input, int remaining) {
+
+    int value = *((unsigned char*) *input);
+
+    /* Replace value with exception if not identical to ISO-8859-1 */
+    if (value >= 0x80 && value <= 0x9F)
+        value = __GUAC_RDP_CP1252_CODEPOINT[value - 0x80];
+
+    (*input)++;
+    return value;
+
+}
+
+int GUAC_READ_ISO8859_1(char** input, int remaining) {
+
+    int value = *((unsigned char*) *input);
+
+    (*input)++;
+    return value;
+
+}
+
 void GUAC_WRITE_UTF8(char** output, int remaining, int value) {
     *output += guac_utf8_write(value, *output, remaining);
 }
@@ -97,3 +157,39 @@ void GUAC_WRITE_UTF16(char** output, int remaining, int value) {
 
 }
 
+void GUAC_WRITE_CP1252(char** output, int remaining, int value) {
+
+    /* If not in ISO-8859-1 part of CP1252, check lookup table */
+    if ((value >= 0x80 && value <= 0x9F) || value > 0xFF) {
+
+        int i;
+        int replacement_value = '?';
+        const int* codepoint = __GUAC_RDP_CP1252_CODEPOINT;
+
+        /* Search lookup table for value */
+        for (i=0x80; i<=0x9F; i++, codepoint++) {
+            if (*codepoint == value) {
+                replacement_value = i;
+                break;
+            }
+        }
+
+        /* Replace value with discovered value (or question mark) */
+        value = replacement_value;
+
+    }
+
+    *((unsigned char*) *output) = (unsigned char) value;
+    (*output)++;
+}
+
+void GUAC_WRITE_ISO8859_1(char** output, int remaining, int value) {
+
+    /* Translate to question mark if out of range */
+    if (value > 0xFF)
+        value = '?';
+
+    *((unsigned char*) *output) = (unsigned char) value;
+    (*output)++;
+}
+
diff --git a/src/common/guac_iconv.h b/src/common/guac_iconv.h
index 98cdce7c..8cc9788a 100644
--- a/src/common/guac_iconv.h
+++ b/src/common/guac_iconv.h
@@ -69,6 +69,16 @@ guac_iconv_read GUAC_READ_UTF8;
  */
 guac_iconv_read GUAC_READ_UTF16;
 
+/**
+ * Read function for CP-1252.
+ */
+guac_iconv_read GUAC_READ_CP1252;
+
+/**
+ * Read function for ISO-8859-1
+ */
+guac_iconv_read GUAC_READ_ISO8859_1;
+
 /**
  * Write function for UTF8.
  */
@@ -79,5 +89,15 @@ guac_iconv_write GUAC_WRITE_UTF8;
  */
 guac_iconv_write GUAC_WRITE_UTF16;
 
+/**
+ * Write function for CP-1252.
+ */
+guac_iconv_write GUAC_WRITE_CP1252;
+
+/**
+ * Write function for ISO-8859-1
+ */
+guac_iconv_write GUAC_WRITE_ISO8859_1;
+
 #endif
 
diff --git a/tests/common/guac_iconv.c b/tests/common/guac_iconv.c
index aca40e0e..cd7f548e 100644
--- a/tests/common/guac_iconv.c
+++ b/tests/common/guac_iconv.c
@@ -71,6 +71,22 @@ void test_guac_iconv() {
         0x00, 0x00
     };
 
+    /* ISO-8859-1 for "papà è bello" */
+    unsigned char test_string_iso8859_1[] = {
+        'p',  'a',  'p', 0xE0, ' ',
+        0xE8, ' ',
+        'b',  'e',  'l', 'l',  'o',
+        0x00
+    };
+
+    /* CP1252 for "papà è bello" */
+    unsigned char test_string_cp1252[] = {
+        'p',  'a',  'p', 0xE0, ' ',
+        0xE8, ' ',
+        'b',  'e',  'l', 'l',  'o',
+        0x00
+    };
+
     /* UTF8 identity */
     test_conversion(
             GUAC_READ_UTF8,  test_string_utf8, sizeof(test_string_utf8),
@@ -91,5 +107,25 @@ void test_guac_iconv() {
             GUAC_READ_UTF16, test_string_utf16, sizeof(test_string_utf16),
             GUAC_WRITE_UTF8, test_string_utf8,  sizeof(test_string_utf8));
 
+    /* UTF16 to ISO-8859-1 */
+    test_conversion(
+            GUAC_READ_UTF16,      test_string_utf16,      sizeof(test_string_utf16),
+            GUAC_WRITE_ISO8859_1, test_string_iso8859_1,  sizeof(test_string_iso8859_1));
+
+    /* UTF16 to CP1252 */
+    test_conversion(
+            GUAC_READ_UTF16,   test_string_utf16,  sizeof(test_string_utf16),
+            GUAC_WRITE_CP1252, test_string_cp1252, sizeof(test_string_cp1252));
+
+    /* CP1252 to UTF8 */
+    test_conversion(
+            GUAC_READ_CP1252, test_string_cp1252, sizeof(test_string_cp1252),
+            GUAC_WRITE_UTF8,  test_string_utf8,   sizeof(test_string_utf8));
+
+    /* ISO-8859-1 to UTF8 */
+    test_conversion(
+            GUAC_READ_ISO8859_1, test_string_iso8859_1, sizeof(test_string_iso8859_1),
+            GUAC_WRITE_UTF8,     test_string_utf8,      sizeof(test_string_utf8));
+
 }