1. Add build script for Java.

2. Add height limitation for code blocks in extra.css.
3. Fix "节点" to "结点".
This commit is contained in:
krahets
2023-02-07 04:43:52 +08:00
parent b14568151c
commit ecbf2d1560
54 changed files with 457 additions and 1633 deletions

View File

@ -1,44 +1,55 @@
"""
File: extract_code_python.py
Created Time: 2023-02-06
Created Time: 2023-02-07
Author: Krahets (krahets@163.com)
"""
import re
import os
import os.path as osp
import glob
class ExtractCodeBlocksPython:
def __init__(self, file_path) -> None:
def __init__(self) -> None:
self.langs = ["python"]
# Pattern to match function names and class names
self.func_pattern = r'(\s*)def\s+(\w+)\s*\('
self.class_pattern = r'class\s+(\w+)'
# Pattern to match the start and end of a block
self.block_end_pattern = '^\s{0,ind}\S+.*\n'
self.block_start_pattern = '^\s{ind}""".+'
def extract(self, file_path):
self.file_path = file_path
with open(file_path) as f:
self.lines = f.readlines()
self.content = "".join(self.lines)
# Regular expression pattern to match function names and class names
self.func_pattern = re.compile(r'(\s*)def\s+(\w+)\s*\(')
self.class_pattern = re.compile(r'class\s+(\w+)')
# Detect and extract all the classes and fucntions
self.classes = self.extract_class_blocks()
self.functions = self.extract_function_blocks()
classes = self.extract_class_blocks()
funcs = self.extract_function_blocks()
return {
"classes": classes,
"funcs": funcs,
}
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)))
# Search the code
for i in range(header_line + 1, len(self.lines)):
if re.search("^\s*\n|^\s{ind}\s+.+\n".replace("ind", str(indentation)),
self.lines[i]) is None:
if re.match(block_end_pattern, self.lines[i]) is not None:
end_line = i
break
# Search the header comment
for i in range(header_line - 1, -1, -1):
if re.search('^\s{ind}""".+'.replace("ind", str(indentation)),
self.lines[i]) is not None:
if re.search(block_start_pattern, self.lines[i]) is not None:
start_line = i
break
func_block = self.lines[start_line:end_line]
@ -55,16 +66,18 @@ class ExtractCodeBlocksPython:
"""
Extract all the functions with given indentation
"""
functions = {}
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 = self.func_pattern.match(self.lines[line_num])
func_match = func_pattern.match(self.lines[line_num])
if func_match is None: continue
# The function should match the input indentation
if len(func_match.group(1)) != indentation: continue
@ -72,9 +85,9 @@ class ExtractCodeBlocksPython:
# Search the block from the header line
start_line, end_line, func_block = self.search_block(header_line, indentation)
# Construct the functions dict
# Construct the funcs dict
func_label = func_match.group(2)
functions[func_label] = {
funcs[func_label] = {
"indentation": indentation,
"line_number": {
"start": start_line,
@ -84,7 +97,7 @@ class ExtractCodeBlocksPython:
"block": func_block,
}
return functions
return funcs
def extract_class_blocks(self):
"""
@ -92,9 +105,11 @@ class ExtractCodeBlocksPython:
"""
classes = {}
class_pattern = re.compile(self.class_pattern)
for line_num, line in enumerate(self.lines):
# Search the class header
class_match = self.class_pattern.match(line)
class_match = class_pattern.match(line)
if class_match is None: continue
header_line = line_num
@ -110,7 +125,7 @@ class ExtractCodeBlocksPython:
"header": header_line,
},
"block": class_block,
"functions": self.extract_function_blocks(
"funcs": self.extract_function_blocks(
indentation=4, start_line=start_line, end_line=end_line)
}