Fixes bug with fortan line cont and preproc

Fixes Preprocessor statements in declarations break parsing #203
Fixes Preprocessor statements in declarations break parsing #4

Author: gnikit

CHANGELOG.md

@@ -38,6 +38,9 @@
   ([gnikit/fortls#16](https://github.com/gnikit/fortls/issues/16))
 - Fixes `END FORALL` end of scope error
   ([gnikit/fortls#18](https://github.com/gnikit/fortls/issues/18))
+- Fixes Fortran line continuation definitions intermingled with preprocessor directives
+  ([#203](https://github.com/hansec/fortran-language-server/issues/203))
+  ([gnikit/fortls#4](https://github.com/gnikit/fortls/issues/4))
 
 ## 1.16.0
 

fortls/parse_fortran.py

@@ -104,6 +104,7 @@
     NAT_VAR_REGEX,
     NON_DEF_REGEX,
     PARAMETER_VAL_REGEX,
+    PP_ANY_REGEX,
     PP_DEF_REGEX,
     PP_INCLUDE_REGEX,
     PP_REGEX,
@@ -978,7 +979,7 @@ def get_code_line(
                     line_ind -= 1
             else:  # Free format file
                 opt_cont_match = FREE_CONT_REGEX.match(curr_line)
-                if opt_cont_match is not None:
+                if opt_cont_match:
                     curr_line = (
                         " " * opt_cont_match.end(0) + curr_line[opt_cont_match.end(0) :]
                     )
@@ -989,7 +990,7 @@ def get_code_line(
                     tmp_no_comm = tmp_line.split("!")[0]
                     cont_ind = tmp_no_comm.rfind("&")
                     opt_cont_match = FREE_CONT_REGEX.match(tmp_no_comm)
-                    if opt_cont_match is not None:
+                    if opt_cont_match:
                         if cont_ind == opt_cont_match.end(0) - 1:
                             break
                         tmp_no_comm = (
@@ -1022,21 +1023,27 @@ def get_code_line(
                 if iComm < 0:
                     iComm = iAmper + 1
                 next_line = ""
+                # Read the next line if needed
                 while (iAmper >= 0) and (iAmper < iComm):
                     if line_ind == line_number + 1:
                         curr_line = curr_line[:iAmper]
                     elif next_line != "":
                         post_lines[-1] = next_line[:iAmper]
                     next_line = self.get_line(line_ind, pp_content)
                     line_ind += 1
+                    # Skip any preprocessor statements when seeking the next line
+                    if PP_ANY_REGEX.match(next_line):
+                        next_line = ""
+                        post_lines.append("")
+                        continue
                     # Skip empty or comment lines
                     match = FREE_COMMENT_LINE_MATCH.match(next_line)
-                    if (next_line.rstrip() == "") or (match is not None):
+                    if next_line.rstrip() == "" or match:
                         next_line = ""
                         post_lines.append("")
                         continue
                     opt_cont_match = FREE_CONT_REGEX.match(next_line)
-                    if opt_cont_match is not None:
+                    if opt_cont_match:
                         next_line = (
                             " " * opt_cont_match.end(0)
                             + next_line[opt_cont_match.end(0) :]
@@ -1056,9 +1063,7 @@ def get_code_line(
     def strip_comment(self, line: str) -> str:
         """Strip comment from line"""
         if self.fixed:
-            if (FIXED_COMMENT_LINE_MATCH.match(line) is not None) and (
-                FIXED_OPENMP_MATCH.match(line) is not None
-            ):
+            if FIXED_COMMENT_LINE_MATCH.match(line) and FIXED_OPENMP_MATCH.match(line):
                 return ""
         else:
             if FREE_OPENMP_MATCH.match(line) is None:

fortls/regex_patterns.py

@@ -111,6 +111,7 @@
 PP_DEF_REGEX = re.compile(r"#(define|undef)[ ]*([\w]+)(\((\w+(,[ ]*)?)+\))?", re.I)
 PP_DEF_TEST_REGEX = re.compile(r"(![ ]*)?defined[ ]*\([ ]*([a-z0-9_]*)[ ]*\)$", re.I)
 PP_INCLUDE_REGEX = re.compile(r"#include[ ]*([\"a-z0-9_\.]*)", re.I)
+PP_ANY_REGEX = re.compile(r"(^#:?\w+)")
 # Context matching rules
 CALL_REGEX = re.compile(r"[ ]*CALL[ ]+[a-z0-9_%]*$", re.I)
 INT_STMNT_REGEX = re.compile(r"^[ ]*[a-z]*$", re.I)

test/test_preproc.py

@@ -34,6 +34,8 @@ def check_return(result_array, checks):
     string += hover_req(file_path, 7, 40)  # multi-lin variable
     string += hover_req(file_path, 8, 7)  # function with if conditional
     string += hover_req(file_path, 9, 7)  # multiline function with if conditional
+    file_path = os.path.join(test_dir, "pp", "preproc_keywords.F90")
+    string += hover_req(file_path, 6, 2)  # ignores PP across Fortran line continuations
     errcode, results = run_request(string, f" --config={root_dir}/.pp_conf.json")
     assert errcode == 0
 
@@ -44,6 +46,7 @@ def check_return(result_array, checks):
         "#define varVar 55",
         "#define ewrite if (priority <= 3) write((priority), format)",
         "#define ewrite2 if (priority <= 3) write((priority), format)",
+        "REAL, CONTIGUOUS, POINTER, DIMENSION(:)",
     )
     assert len(ref_results) == len(results) - 1
     check_return(results[1:], ref_results)

test/test_server.py

@@ -186,6 +186,7 @@ def check_return(result_array):
             ["test_int", 2, 0],
             ["test_mod", 2, 0],
             ["test_nonint_mod", 2, 0],
+            ["test_preproc_keywords", 2, 0],
             ["test_program", 2, 0],
             ["test_rename_sub", 6, 9],
             ["test_select", 2, 0],

test/test_source/pp/.pp_conf.json

@@ -6,5 +6,6 @@
   "enable_code_actions": true,
   "pp_suffixes": [".h", ".F90"],
   "incl_suffixes": [".h"],
-  "include_dirs": ["include"]
+  "include_dirs": ["include"],
+  "pp_defs": { "HAVE_CONTIGUOUS": "" }
 }

test/test_source/pp/preproc_keywords.F90

@@ -0,0 +1,10 @@
+program test_preproc_keywords
+REAL &
+#ifdef HAVE_CONTIGUOUS
+, CONTIGUOUS &
+#endif
+, POINTER :: & 
+var1(:), &
+var2(:)
+
+end program test_preproc_keywords