2
0
mirror of https://gitlab.com/ita1024/waf.git synced 2024-12-11 11:20:21 +01:00
waf/waflib/Tools/fc_scan.py

122 lines
3.0 KiB
Python
Raw Normal View History

2011-09-10 11:13:51 +02:00
#! /usr/bin/env python
# encoding: utf-8
# DC 2008
# Thomas Nagy 2010 (ita)
import re
from waflib import Utils, Task, TaskGen, Logs
from waflib.TaskGen import feature, before_method, after_method, extension
from waflib.Configure import conf
2013-12-20 18:00:05 +01:00
INC_REGEX = """(?:^|['">]\s*;)\s*(?:|#\s*)INCLUDE\s+(?:\w+_)?[<"'](.+?)(?=["'>])"""
2011-09-10 11:13:51 +02:00
USE_REGEX = """(?:^|;)\s*USE(?:\s+|(?:(?:\s*,\s*(?:NON_)?INTRINSIC)?\s*::))\s*(\w+)"""
MOD_REGEX = """(?:^|;)\s*MODULE(?!\s*PROCEDURE)(?:\s+|(?:(?:\s*,\s*(?:NON_)?INTRINSIC)?\s*::))\s*(\w+)"""
re_inc = re.compile(INC_REGEX, re.I)
re_use = re.compile(USE_REGEX, re.I)
re_mod = re.compile(MOD_REGEX, re.I)
class fortran_parser(object):
"""
This parser will return:
* the nodes corresponding to the module names that will be produced
* the nodes corresponding to the include files used
* the module names used by the fortran file
"""
def __init__(self, incpaths):
self.seen = []
"""Files already parsed"""
self.nodes = []
"""List of :py:class:`waflib.Node.Node` representing the dependencies to return"""
self.names = []
"""List of module names to return"""
self.incpaths = incpaths
"""List of :py:class:`waflib.Node.Node` representing the include paths"""
def find_deps(self, node):
"""
Parse a fortran file to read the dependencies used and provided
:param node: fortran file to read
:type node: :py:class:`waflib.Node.Node`
:return: lists representing the includes, the modules used, and the modules created by a fortran file
:rtype: tuple of list of strings
"""
txt = node.read()
incs = []
uses = []
mods = []
for line in txt.splitlines():
# line by line regexp search? optimize?
m = re_inc.search(line)
if m:
incs.append(m.group(1))
m = re_use.search(line)
if m:
uses.append(m.group(1))
m = re_mod.search(line)
if m:
mods.append(m.group(1))
return (incs, uses, mods)
def start(self, node):
"""
Start the parsing. Use the stack self.waiting to hold the nodes to iterate on
:param node: fortran file
:type node: :py:class:`waflib.Node.Node`
"""
self.waiting = [node]
while self.waiting:
nd = self.waiting.pop(0)
self.iter(nd)
def iter(self, node):
"""
Process a single file in the search for dependencies, extract the files used
the modules used, and the modules provided.
"""
path = node.abspath()
incs, uses, mods = self.find_deps(node)
for x in incs:
if x in self.seen:
continue
self.seen.append(x)
self.tryfind_header(x)
for x in uses:
name = "USE@%s" % x
if not name in self.names:
self.names.append(name)
for x in mods:
name = "MOD@%s" % x
if not name in self.names:
self.names.append(name)
def tryfind_header(self, filename):
"""
Try to find an include and add it the nodes to process
:param filename: file name
:type filename: string
"""
found = None
for n in self.incpaths:
found = n.find_resource(filename)
if found:
self.nodes.append(found)
self.waiting.append(found)
break
if not found:
if not filename in self.names:
self.names.append(filename)