exclude Composition_Exclusion and fix range parser

Author: tamaroning

gcc/rust/util/make-rust-unicode.py

@@ -57,18 +57,38 @@
 nfc_qc_maybe_ranges = []
 numeric_codepoints = []
 
+# Note that an element of range `[m, n]` (a list in python) represents [m, n)
+
+
+def binary_search_ranges(ranges, target):
+    low = 0
+    high = len(ranges) - 1
+    while low <= high:
+        mid = (low + high) // 2
+        start, end = ranges[mid]
+        if start <= target <= end - 1:
+            return mid  # target found. returns index.
+        elif target < start:
+            high = mid - 1
+        else:
+            low = mid + 1
+    # target not found.
+    return -1
+
 
 # Utility function to parse '<codepoint>...<codepoint>' or '<codepoint>'
 def parse_codepoint_range(range_str):
     codepoint_range = range_str.split("..")
     assert len(codepoint_range) == 1 or len(codepoint_range) == 2, "Invalid format"
     start_cp, end_cp = 0, 0
     if len(codepoint_range) == 1:
+        # m..n => [m, n+1)
         start_cp = int(codepoint_range[0], 16)
         end_cp = start_cp + 1
     else:
+        # m => [m, m+1)
         start_cp = int(codepoint_range[0], 16)
-        end_cp = int(codepoint_range[1], 16)
+        end_cp = int(codepoint_range[1], 16) + 1
     return [start_cp, end_cp]
 
 
@@ -170,6 +190,24 @@ def write_decomposition():
     print("};")
 
 
+def write_recomposition():
+    # non const.
+    print("std::map<std::pair<uint32_t, uint32_t>, uint32_t> RECOMPOSITION_MAP = {{")
+    print("  // clang-format off")
+    for cp in decomposition_map:
+        if binary_search_ranges(composition_exclusion_ranges, cp) != -1:
+            continue
+        if len(decomposition_map[cp]) == 1:
+            d1 = decomposition_map[cp][0]
+            d2 = 0
+        else:
+            d1 = decomposition_map[cp][0]
+            d2 = decomposition_map[cp][1]
+        print("  {{{{{:#06x}, {:#06x}}}, {:#06x}}},".format(d1, d2, cp))
+    print("  // clang-format on")
+    print("}};")
+
+
 def write_ccc():
     print("const std::map<uint32_t, int32_t> CCC_TABLE = {")
     print("  // clang-format off")
@@ -179,19 +217,6 @@ def write_ccc():
     print("};")
 
 
-# TODO: remove this
-# def write_composition_exclusion():
-#    print(
-#        "const std::array<std::pair<uint32_t, uint32_t>, {}>".format(
-#            len(composition_exclusion_ranges)
-#        )
-#    )
-#    print("NO_RECOMP_RANGES = {{")
-#    for r in composition_exclusion_ranges:
-#        print("  {{{:#06x}, {:#06x}}},".format(r[0], r[1]))
-#    print("}};")
-
-
 def write_alphabetic():
     print(
         "const std::array<std::pair<uint32_t, uint32_t>, {}> ALPHABETIC_RANGES = {{{{".format(
@@ -224,23 +249,6 @@ def write_numeric():
     print("}};")
 
 
-def write_recomposition():
-    # non const.
-    # TODO: Exclude `Composition_Exclusion`s
-    print("std::map<std::pair<uint32_t, uint32_t>, uint32_t> RECOMPOSITION_MAP = {{")
-    print("  // clang-format off")
-    for cp in decomposition_map:
-        if len(decomposition_map[cp]) == 1:
-            d1 = decomposition_map[cp][0]
-            d2 = 0
-        else:
-            d1 = decomposition_map[cp][0]
-            d2 = decomposition_map[cp][1]
-        print("  {{{{{:#06x}, {:#06x}}}, {:#06x}}},".format(d1, d2, cp))
-    print("  // clang-format on")
-    print("}};")
-
-
 def main():
     if len(sys.argv) != 4:
         print("too few arguments", file=sys.stderr)