Changeset 4735

Timestamp:

Aug 21, 2015, 1:44:51 PM (4 years ago)

Author:

cameron

Message:

Support for Grapheme, Sentence, Word break properties

Files:

• icGREP/icgrep-devel/icgrep/UCD/GraphemeBreakProperty.h (added)
• icGREP/icgrep-devel/icgrep/UCD/PropertyObjectTable.h (modified) (4 diffs)
• icGREP/icgrep-devel/icgrep/UCD/SentenceBreakProperty.h (added)
• icGREP/icgrep-devel/icgrep/UCD/WordBreakProperty.h (added)
• icGREP/icgrep-devel/icgrep/UCD/resolve_properties.cpp (modified) (2 diffs)
• proto/charsetcompiler/UCD/UCD_properties.py (modified) (2 diffs)

icGREP/icgrep-devel/icgrep/UCD/PropertyObjectTable.h

@@ -16 +16 @@
 #include "DerivedGeneralCategory.h"
 #include "EastAsianWidth.h"
+#include "GraphemeBreakProperty.h"
 #include "HangulSyllableType.h"
 #include "LineBreak.h"
@@ -21 +22 @@
 #include "ScriptExtensions.h"
 #include "Scripts.h"
+#include "SentenceBreakProperty.h"
+#include "WordBreakProperty.h"
 
 namespace UCD {
@@ -69 +72 @@
     &EA_ns::property_object,
     &GC_ns::property_object,
-    new UnsupportedPropertyObject(GCB, PropertyObject::ClassTypeId::EnumeratedProperty),
+    &GCB_ns::property_object,
     &HST_ns::property_object,
     new UnsupportedPropertyObject(InPC, PropertyObject::ClassTypeId::EnumeratedProperty),
@@ -81 +84 @@
     new UnsupportedPropertyObject(NFKD_QC, PropertyObject::ClassTypeId::EnumeratedProperty),
     new UnsupportedPropertyObject(nt, PropertyObject::ClassTypeId::EnumeratedProperty),
-    new UnsupportedPropertyObject(SB, PropertyObject::ClassTypeId::EnumeratedProperty),
-    new UnsupportedPropertyObject(WB, PropertyObject::ClassTypeId::EnumeratedProperty),
+    &SB_ns::property_object,
+    &WB_ns::property_object,
     &AHEX_ns::property_object,
     &ALPHA_ns::property_object,

icGREP/icgrep-devel/icgrep/UCD/resolve_properties.cpp

@@ -77 +77 @@
         property->setFunctionName("__get_blk_" + BLK_ns::enum_names[valcode]);
     }
+    else if (theprop == GCB) {
+        // Grapheme Cluster Break property identified
+        int valcode = GetPropertyValueEnumCode(GCB, value);
+        if (valcode < 0) {
+             throw UnicodePropertyExpressionError("Erroneous property value for grapheme cluster break property");
+        }
+        property->setFunctionName("__get_gcb_" + GCB_ns::enum_names[valcode]);
+    }
+    else if (theprop == WB) {
+        // Word Break property identified
+        int valcode = GetPropertyValueEnumCode(WB, value);
+        if (valcode < 0) {
+             throw UnicodePropertyExpressionError("Erroneous property value for word break property");
+        }
+        property->setFunctionName("__get_wb_" + WB_ns::enum_names[valcode]);
+    }
+    else if (theprop == lb) {
+        // Line Break property identified
+        int valcode = GetPropertyValueEnumCode(lb, value);
+        if (valcode < 0) {
+             throw UnicodePropertyExpressionError("Erroneous property value for line break property");
+        }
+        property->setFunctionName("__get_lb_" + LB_ns::enum_names[valcode]);
+    }
+    else if (theprop == SB) {
+        // Sentence Break property identified
+        int valcode = GetPropertyValueEnumCode(SB, value);
+        if (valcode < 0) {
+             throw UnicodePropertyExpressionError("Erroneous property value for sentence break property");
+        }
+        property->setFunctionName("__get_lb_" + SB_ns::enum_names[valcode]);
+    }
     else if (isa<BinaryPropertyObject>(property_object_table[theprop])){
         auto valit = Binary_ns::aliases_only_map.find(value);
@@ -187 +219 @@
             }
             auto theprop = propit->second;
-            if (theprop == gc) {
-                // General Category
-                return cast<EnumeratedPropertyObject>(property_object_table[gc])->GetCodepointSet(value);
-            }
-            else if (theprop == sc) {
-                // Script property identified
-                return cast<EnumeratedPropertyObject>(property_object_table[sc])->GetCodepointSet(value);
-            }
-            else if (theprop == scx) {
-                // Script extension property identified
-                return cast<ExtensionPropertyObject>(property_object_table[scx])->GetCodepointSet(value);
-            }
-            else if (theprop == blk) {
-                // Block property identified
-                return cast<EnumeratedPropertyObject>(property_object_table[blk])->GetCodepointSet(value);
+            if (EnumeratedPropertyObject * p = dyn_cast<EnumeratedPropertyObject>(property_object_table[theprop])){
+                return p->GetCodepointSet(value);
             }
             else if (BinaryPropertyObject * p = dyn_cast<BinaryPropertyObject>(property_object_table[theprop])){

proto/charsetcompiler/UCD/UCD_properties.py

@@ -14 +14 @@
 from UCD_parser import *
 
-UCD_dir = "7.0.0"
+UCD_dir = "8.0.0"
 
 PropertyAliases_template = r"""
@@ -240 +240 @@
     ucd.generate_property_value_file('LineBreak', 'lb')
     #
+    # Grapheme Cluster Break property
+    ucd.generate_property_value_file('auxiliary/GraphemeBreakProperty', 'GCB')
+    #
+    # Sentence Break property
+    ucd.generate_property_value_file('auxiliary/SentenceBreakProperty', 'SB')
+    #
+    # Word Break property
+    ucd.generate_property_value_file('auxiliary/WordBreakProperty', 'WB')
+    #
     # East Asian Width
     ucd.generate_property_value_file('EastAsianWidth', 'ea')