algorithm/hello-algo
1
0
Fork 0
mirror of https://github.com/krahets/hello-algo.git synced 2025-07-20 21:52:30 +08:00
Code Issues Packages Projects Releases Wiki Activity

Update code extractor of Java.

Browse Source
This commit is contained in:
krahets
2023-02-07 19:05:19 +08:00
parent 95bffcfb36
commit 8f512c2840
9 changed files with 99 additions and 87 deletions
Download Patch File Download Diff File Expand all files Collapse all files

100
docs/utils/extract_code_java.py
View File

@ -8,8 +8,9 @@ import re
import os
import os.path as osp
class ExtractCodeBlocksJava:
def __init__(self) -> None:
def __init__(self) -> None:
self.langs = ["java"]
# Pattern to match function names and class names
self.func_pattern = r'(\s+)(public|private|)\s*(static|)\s*(\S+)\s+(\w+)(\(.*\))\s+\{'
@ -17,40 +18,38 @@ class ExtractCodeBlocksJava:
# Pattern to match the start and end of a block
self.block_end_pattern = '^\s{ind}\}'
self.block_start_pattern = '^\s{ind}\/\*.+\*\/'
def extract(self, file_path):
"""
Extract classes and functions from a markdown document
"""
self.file_path = file_path
with open(file_path) as f:
self.lines = f.readlines()
self.content = "".join(self.lines)
# Detect and extract all the classes along with its fucntions
# Detect and extract all the classes and fucntions
classes = self.extract_class_blocks()
# Remove 'static'
self.post_process(classes)
funcs = self.extract_function_blocks()
self.post_process(classes, funcs)
return {
"classes": classes
"classes": classes,
"funcs": funcs,
}
def post_process(self, classes):
for clas in classes.values():
funcs = clas["funcs"]
for func in funcs.values():
for i, line in enumerate(func["block"]):
if "static " in line:
func["block"][i] = line.replace("static ", "")
break
def search_block(self, header_line, indentation):
"""
Search class/function block given the header_line and indentation
"""
start_line, end_line = 0, len(self.lines)
block_end_pattern = re.compile(self.block_end_pattern.replace("ind", str(indentation)))
block_start_pattern = re.compile(self.block_start_pattern.replace("ind", str(indentation)))
block_end_pattern = re.compile(
self.block_end_pattern.replace("ind", str(indentation)))
block_start_pattern = re.compile(
self.block_start_pattern.replace("ind", str(indentation)))
# Search the code
for i in range(header_line + 1, len(self.lines)):
if re.match(block_end_pattern, self.lines[i]) is not None:
@ -61,39 +60,35 @@ class ExtractCodeBlocksJava:
if re.search(block_start_pattern, self.lines[i]) is not None:
start_line = i
break
code_block = self.lines[start_line:end_line + 1]
# Remove empty lines at bottom
for i in range(len(code_block) - 1, -1, -1):
if re.search("^\s*\n", code_block[i]) is None:
break
end_line -= 1
return start_line, end_line, self.lines[start_line:end_line + 1]
def extract_function_blocks(self, indentation=4, start_line=-1, end_line=-1):
"""
Extract all the functions with given indentation
"""
funcs = {}
if start_line == -1:
start_line = 0
if end_line == -1:
end_line = len(self.lines) - 1
func_pattern = re.compile(self.func_pattern)
for line_num in range(start_line, end_line + 1):
# Search the function header
func_match = func_pattern.match(self.lines[line_num])
if func_match is None: continue
if func_match is None:
continue
# The function should match the input indentation
if len(func_match.group(1)) != indentation: continue
if len(func_match.group(1)) != indentation:
continue
header_line = line_num
# Search the block from the header line
start_line, end_line, func_block = self.search_block(header_line, indentation)
start_line, end_line, func_block = self.search_block(
header_line, indentation)
# Construct the funcs dict
func_label = func_match.group(5)
funcs[func_label] = {
@ -105,7 +100,7 @@ class ExtractCodeBlocksJava:
},
"block": func_block,
}
return funcs
def extract_class_blocks(self):
@ -113,17 +108,19 @@ class ExtractCodeBlocksJava:
Extract all the classes with given indentation
"""
classes = {}
class_pattern = re.compile(self.class_pattern)
for line_num, line in enumerate(self.lines):
# Search the class header
class_match = class_pattern.match(line)
if class_match is None: continue
if class_match is None:
continue
header_line = line_num
# Search the block from the header line
start_line, end_line, class_block = self.search_block(header_line, 0)
start_line, end_line, class_block = self.search_block(
header_line, 0)
# Construct the classes dict
class_label = class_match.group(2)
classes[class_label] = {
@ -137,9 +134,24 @@ class ExtractCodeBlocksJava:
"funcs": self.extract_function_blocks(
indentation=4, start_line=start_line, end_line=end_line)
}
return classes
def post_process(self, classes, funcs):
"""
Process the classes and functions
"""
def remove_keyword(func):
block = func["block"]
header_line = func["line_number"]["header"] - \
func["line_number"]["start"]
block[header_line] = block[header_line] \
.replace("static ", "").replace("public ", "").replace("private ", "")
for clas in classes.values():
for func in clas["funcs"].values():
remove_keyword(func)
for func in funcs.values():
remove_keyword(func)
# ext = ExtractCodeBlocksJava()
# ext.extract("codes/java/chapter_array_and_linkedlist/array.java")
# ext.extract("codes/java/chapter_array_and_linkedlist/my_list.java")