feat: SQL between/hostvar-key alignment, class-condition parsing, gcov merge across scenario runs

cobol_testgen/read.py

@@ -207,13 +207,13 @@ _RE_SQL_INC = re.compile(
 _BUILTIN_SQLCA = """\
         01 SQLCA.
             05 SQLCAID      PIC X(8).
-            05 SQLCABC      PIC S9(9) COMP.
-            05 SQLCODE      PIC S9(9) COMP.
+            05 SQLCABC      PIC S9(9) COMP-5.
+            05 SQLCODE      PIC S9(9) COMP-5.
             05 SQLERRM.
-                10 SQLERRML PIC S9(4) COMP.
+                10 SQLERRML PIC S9(4) COMP-5.
                 10 SQLERRMC PIC X(70).
             05 SQLERRP      PIC X(8).
-            05 SQLERRD      OCCURS 6 TIMES PIC S9(9) COMP.
+            05 SQLERRD      OCCURS 6 TIMES PIC S9(9) COMP-5.
             05 SQLWARN.
                 10 SQLWARN0 PIC X.
                 10 SQLWARN1 PIC X.
@@ -523,6 +523,11 @@ def parse_pic(pic_str: str) -> PicInfo:
     elif expanded[0] == 'A':
         info.type = 'alphabetic'
         info.length = len(expanded)
+    elif expanded[0] == 'N':
+        # PIC N(n) = national/DBCS: 每个字符占 2 字节（GnuCOBOL 实际存储）。
+        # 按 n 字节计算会导致 FD 记录布局/偏移错位（如 N(040)=80B 却按 40B 写）。
+        info.type = 'national'
+        info.length = len(expanded) * 2
     elif expanded[0] in ('Z', '*', '$', '+', '-'):
         info.type = 'numeric-edited'
         info.digits = expanded.count('9')
@@ -597,15 +602,17 @@ def parse_file_control(source: str) -> dict:
     result = {}
     for sel_m in re.finditer(
         r'SELECT\s+(\w[\w-]*)\s+[^.]*?\bASSIGN\s+TO\s+'
-        r'(?:(["\'])(.*?)\2|(\w[\w-]*))'
+        r'(?:(["\'])(.*?)\2|EXTERNAL\s+(\w[\w-]*)|(\w[\w-]*))'
         r'[^.]*\.',
         fc, re.IGNORECASE
     ):
         name = sel_m.group(1).upper()
         if sel_m.group(2):
             assign_to = sel_m.group(3).upper()
-        else:
+        elif sel_m.group(4):
             assign_to = sel_m.group(4).upper()
+        else:
+            assign_to = sel_m.group(5).upper()
         clause = sel_m.group(0)
         org_m = re.search(r'ORGANIZATION\s+(LINE\s+)?SEQUENTIAL', clause, re.IGNORECASE)
         if org_m and org_m.group(1):
@@ -673,3 +680,57 @@ def scan_open_statements(source: str) -> dict:
                 for fname in re.findall(r'\w[\w-]*', seg_m.group(2)):
                     dirs[fname.upper()] = direction
     return dirs
+
+
+def scan_sort_merge_directions(source: str) -> dict:
+    """Parse MERGE / SORT statements, returns {file_name: 'INPUT'|'OUTPUT'}.
+
+    MERGE/SORT 的 USING 文件不会被 OPEN（由语句隐式打开），因此
+    scan_open_statements 无法识别。此函数扫描 PROCEDURE DIVISION 中
+    MERGE ... USING <f1> <f2> 与 SORT ... USING <f> GIVING <g> 子句，
+    将 USING 文件识别为 INPUT、GIVING 文件识别为 OUTPUT。
+
+    兼容性：对任何含 MERGE/SORT 的程序通用，非 MERGE/SORT 程序返回空 dict。
+    """
+    dirs = {}
+    _KEYWORDS = {'USING', 'GIVING', 'PROCEDURE', 'KEY', 'ASCENDING', 'DESCENDING', 'ON', 'OUTPUT', 'INPUT'}
+    for m in re.finditer(
+        r'\b(MERGE|SORT)\s+\w[\w-]*\s+.*?\b(?:USING|GIVING)\s+(.+?)\s*(?:OUTPUT|INPUT)?\s*PROCEDURE\s+\w[\w-]*\s*\.',
+        source, re.IGNORECASE | re.DOTALL
+    ):
+        stmt = m.group(0)
+        kind = m.group(1).upper()
+        files_raw = m.group(2)
+        files = re.findall(r'\b(\w[\w-]*)\b', files_raw)
+        for fn in files:
+            fn = fn.upper()
+            if fn not in _KEYWORDS:
+                dirs[fn] = 'INPUT'
+        if kind == 'SORT':
+            g = re.search(r'\bGIVING\s+(\w[\w-]*)', stmt, re.IGNORECASE)
+            if g:
+                dirs[g.group(1).upper()] = 'OUTPUT'
+    # 单行回退：无 PROCEDURE 的 MERGE/SORT（如 MERGE F ON KEY K USING A B.）
+    for m in re.finditer(
+        r'\b(MERGE|SORT)\s+\w[\w-]*\s+.*?\b(?:USING|GIVING)\s+(.+?)\.',
+        source, re.IGNORECASE
+    ):
+        kind = m.group(1).upper()
+        files_raw = m.group(2)
+        files = re.findall(r'\b(\w[\w-]*)\b', files_raw)
+        for fn in files:
+            fn = fn.upper()
+            if fn not in _KEYWORDS:
+                dirs[fn] = 'INPUT'
+        if kind == 'SORT':
+            g = re.search(r'\bGIVING\s+(\w[\w-]*)', m.group(0), re.IGNORECASE)
+            if g:
+                dirs[g.group(1).upper()] = 'OUTPUT'
+    return dirs
+
+
+def scan_all_file_directions(source: str) -> dict:
+    """OPEN 方向 + MERGE/SORT USING/GIVING 方向的合并（输入方向全集）。"""
+    dirs = scan_open_statements(source)
+    dirs.update(scan_sort_merge_directions(source))
+    return dirs