Merge pull request #13 from keyhr/master

Fixed line break inserting

c_formatter_42/data/.clang-format

@@ -92,7 +92,7 @@ BreakBeforeTernaryOperators: true
 
 # ColumnLimit (unsigned)
 # The column limit.
-ColumnLimit: 80
+ColumnLimit: 0
 
 
 # FixNamespaceComments (bool)

c_formatter_42/formatters/line_breaker.py

@@ -1,30 +1,32 @@
 import re
+from c_formatter_42.formatters import helper
 
 
 def line_breaker(content: str, column_limit: int = 80) -> str:
     lines = content.split("\n")
     lines = list(map(lambda s: insert_break(s, column_limit), lines))
+
     return "\n".join(lines)
 
 
 def insert_break(line: str, column_limit: int) -> str:
     if line_length(line) <= column_limit:
         return line
 
-    line_indent_level = indent_level(line)
-    tabulation = "\t" * (line_indent_level + 1)
-
     # break at all breakable spaces (space after comma or space before binary operators)
     breakable_space_pattern = r"((?<=,) | (?=[+\-*/%])(?!\*+\S|\+\+|\-\-))"
     line = re.sub(breakable_space_pattern, "\n", line)
     segments = line.split("\n")
 
+    line_indent_level = indent_level(line)
+
     # join as many segments as it doesn't exceed line length limit
     line = segments[0]
     current_line_length = line_length(segments[0])
     for segment in segments[1:]:
         current_line_length += line_length(segment) + 1
         if current_line_length > column_limit:
+            tabulation = "\t" * (line_indent_level + additional_indent_level(line))
             line = ("\n" + tabulation).join([line, segment])
             current_line_length = line_length(tabulation + segment)
         else:
@@ -33,15 +35,52 @@ def insert_break(line: str, column_limit: int) -> str:
     return line
 
 
+#
+# additional indent level increases in proportion to corresponds paren depth
+#
+# (examples)
+# foo() * bar() * baz()
+# ~~~~~~~~~~~~~^~~~~~~~ when line breaks here,
+# foo() * bar()
+# >   * baz()           next line should be indented with 1 tab
+# ===
+# (foo(bar() * baz()))
+# ~~~~~~~~~~^~~~~~~~~    when line breaks here,
+# (foo(bar()
+# >   >   >   * baz()))  next line should be indented with 3 tabs
+#
+# function declaration, user defined type (?), and control statement needs one less tab (discount)
+#
+def additional_indent_level(s: str) -> int:
+    additional_indent_level = 1
+
+    discount_pattern = r"(^\t*{type}\t+.*?[a-zA-Z0-9_]\()|(^\t*typedef)|(^\t*(if|while))"
+    discount_pattern = discount_pattern.format(
+        type=helper.REGEX_TYPE,
+    )
+    if re.match(discount_pattern, s):
+        additional_indent_level = 0
+
+    is_surrounded_sq = False
+    is_surrounded_dq = False
+
+    for c in s:
+        if c == "'":
+            is_surrounded_sq = not is_surrounded_sq
+        elif c == '"':
+            is_surrounded_dq = not is_surrounded_dq
+        elif c == "(" and not is_surrounded_sq and not is_surrounded_dq:
+            additional_indent_level += 1
+        elif c == ")" and not is_surrounded_sq and not is_surrounded_dq:
+            additional_indent_level -= 1
+
+    return additional_indent_level
+
+
 def line_length(line: str) -> int:
     line = line.expandtabs(4)
     return len(line)
 
 
 def indent_level(line: str) -> int:
-    last_tab_occurance = line.rfind("\t")
-    if last_tab_occurance < 0:
-        return 0
-
-    line = line[: (last_tab_occurance + 1)]
-    return int(len(line.expandtabs(4)) / 4)
+    return line.count("\t")

setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = c_formatter_42
-version = 0.1.2
+version = 0.1.3
 description = formatting tool complient with 42 school's norm
 long_description = file: README.md
 long_description_content_type = text/markdown

tests/formatters/test_clang_format.py

@@ -33,26 +33,26 @@ def test_clang_format_empty():
     assert clang_format("") == ""
 
 
-def test_clang_format_dont_join_lines():
-    input = """
-if (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
-\t|| bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb
-\t|| cccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc)
-"""
-    output = """
-if (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa ||
-\tbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb ||
-\tcccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc)
-"""
-    assert clang_format(input) == output
-
-    input = """
-if (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa ||
-\tbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb ||
-\tcccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc)
-"""
-    assert clang_format(input) == input
-
+# def test_clang_format_dont_join_lines():
+#     input = """
+# if (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
+# \t|| bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb
+# \t|| cccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc)
+# """
+#     output = """
+# if (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa ||
+# \tbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb ||
+# \tcccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc)
+# """
+#     assert clang_format(input) == output
+# 
+#     input = """
+# if (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa ||
+# \tbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb ||
+# \tcccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc)
+# """
+#     assert clang_format(input) == input
+# 
 
 def test_clang_format_non_array_assignment_packing():
     input = """
@@ -71,19 +71,6 @@ def test_clang_format_non_array_assignment_packing():
     assert clang_format(input) == input
 
 
-def test_clang_format_long_function_declaration():
-    input = """
-static void st_merge_fields_in_curr(
-\t\tchar *strs[3], t_tok_lst **curr, t_tok_lst *fields)
-"""
-    output = """
-static void st_merge_fields_in_curr(char *strs[3],
-\t\t\t\t\t\t\t\t\tt_tok_lst **curr,
-\t\t\t\t\t\t\t\t\tt_tok_lst *fields)
-"""
-    assert clang_format(input) == output
-
-
 @contextmanager
 def change_temp_dir_context():
     tempdir = tempfile.mkdtemp("c_formatter_42")

tests/formatters/test_hoist.py

@@ -130,7 +130,20 @@ def test_hoist_ex():
 """)
 
 # TODO test on weird types
+    output = """\
+void foo()
+{
+\tt_list\t*bar;
 
+\tbar = (t_list *)malloc(sizeof(t_list) * (count_elements(baz) + 1));
+}
+"""
+    assert output == hoist("""\
+void foo()
+{
+\tt_list\t*bar = (t_list *)malloc(sizeof(t_list) * (count_elements(baz) + 1));
+}
+""")
 
 @pytest.mark.skip()
 def test_hoist_array():

tests/formatters/test_line_breaker.py

@@ -17,7 +17,7 @@ def test_line_indent_depth_basic_3():
     input = """\
 \t\t\t + 2 + 2 + 2\t
 """
-    assert 7 == indent_level(input)
+    assert 4 == indent_level(input)
 
 
 def test_insert_line_break_basic_1():
@@ -40,8 +40,8 @@ def test_insert_line_break_basic_2():
 def test_insert_line_break_basic_3():
     output = """\
 \t\t\t\treturn (fooooooooooooooooooooooooo(a, b, cccccccccccc,
-\t\t\t\t\tddddddddddddd, eeeeeeeeeeeeeeee, fffffffffffffff,
-\t\t\t\t\tgggggggggggg, hhhhhhhhhhhhhhhhhh));
+\t\t\t\t\t\t\tddddddddddddd, eeeeeeeeeeeeeeee, fffffffffffffff,
+\t\t\t\t\t\t\tgggggggggggg, hhhhhhhhhhhhhhhhhh));
 """
     assert output == line_breaker("""\
 \t\t\t\treturn (fooooooooooooooooooooooooo(a, b, cccccccccccc, ddddddddddddd, eeeeeeeeeeeeeeee, fffffffffffffff, gggggggggggg, hhhhhhhhhhhhhhhhhh));
@@ -50,7 +50,7 @@ def test_insert_line_break_basic_3():
 def test_insert_line_break_basic_4():
     output = """\
 void\t\tlooooooooooooooooooooooong = 2 + 2 + 2 + 2 + 2 + 2 + 2 + 2 + 2 + 2
-\t\t\t\t+ 2 + 2 + 2;
+\t\t\t+ 2 + 2 + 2;
 """
     assert output == line_breaker("""\
 void\t\tlooooooooooooooooooooooong = 2 + 2 + 2 + 2 + 2 + 2 + 2 + 2 + 2 + 2 + 2 + 2 + 2;
@@ -151,3 +151,37 @@ def test_insert_line_break_basic_22():
     output = ",\n\taaaa* b"
     assert output == line_breaker(", aaaa* b", 7)
 
+def test_insert_line_break_basic_23():
+    output = "foooooo(bar\n\t\t* baz)"
+    assert output == line_breaker("foooooo(bar * baz)", 7)
+
+
+def test_insert_line_break_long_function_declaration():
+    input = """
+static void\tst_merge_fields_in_curr(char *strs[3], t_tok_lst **curr, t_tok_lst *fields)
+"""
+    output = """
+static void\tst_merge_fields_in_curr(char *strs[3], t_tok_lst **curr,
+\t\tt_tok_lst *fields)
+"""
+    assert line_breaker(input) == output
+
+def test_insert_line_break_control_statement_1():
+    input = """\
+\twhile (true + false)
+"""
+    output = """\
+\twhile (true
+\t\t+ false)
+"""
+    assert line_breaker(input, 7) == output
+
+def test_insert_line_break_control_statement_2():
+    input = """\
+\tif (true + false)
+"""
+    output = """\
+\tif (true
+\t\t+ false)
+"""
+    assert line_breaker(input, 5) == output