tempgen.parsers.xlsx

View Source

import os
import xml.etree.ElementTree as ET
import shutil
from tempgen.utils import extract_zip, make_zip
from tempgen.parsers.parser import AbstractParser

class Parser(AbstractParser):
    def parse(self, path, container, parse_entry, find_matches):
        temp_path = path + '_temp'
        extract_zip(path, temp_path)
        ET.register_namespace('','http://schemas.openxmlformats.org/spreadsheetml/2006/main')
        tree = ET.parse(os.sep.join([temp_path, 'xl', 'sharedStrings.xml']))
        root = tree.getroot()
        for si in root:
            for t in si:
                text = t.text
                matches = find_matches(str(text))
                for match in matches:
                    payload = parse_entry(match, path)
                    container[payload['id']] = payload ## TODO: add check here for duplicates
        shutil.rmtree(temp_path)

    def replace(self, source_path, target_path, compute_match, replacements, update_external = False):
        temp_path = source_path + '_temp'
        try:
            shutil.rmtree(temp_path)
        except Exception as e:
            pass
        extract_zip(source_path, temp_path)
            
        ET.register_namespace('','http://schemas.openxmlformats.org/spreadsheetml/2006/main')
        tree = ET.parse(os.sep.join([temp_path, 'xl', 'sharedStrings.xml']))
        root = tree.getroot()
        for si in root:
            for t in si:
                text = t.text
                local_to_replace = compute_match(t.text, {}, replacements, source_path, update_external)
                for match, value in local_to_replace.items():
                    text = text.replace(match, value)
                t.text = text
        tree.write(os.path.join(temp_path, 'xl', 'sharedStrings.xml'))
        make_zip(temp_path, target_path)
        shutil.rmtree(temp_path)

# class Parser(tempgen.parsers.parser.AbstractParser):

View Source

class Parser(AbstractParser):
    def parse(self, path, container, parse_entry, find_matches):
        temp_path = path + '_temp'
        extract_zip(path, temp_path)
        ET.register_namespace('','http://schemas.openxmlformats.org/spreadsheetml/2006/main')
        tree = ET.parse(os.sep.join([temp_path, 'xl', 'sharedStrings.xml']))
        root = tree.getroot()
        for si in root:
            for t in si:
                text = t.text
                matches = find_matches(str(text))
                for match in matches:
                    payload = parse_entry(match, path)
                    container[payload['id']] = payload ## TODO: add check here for duplicates
        shutil.rmtree(temp_path)

    def replace(self, source_path, target_path, compute_match, replacements, update_external = False):
        temp_path = source_path + '_temp'
        try:
            shutil.rmtree(temp_path)
        except Exception as e:
            pass
        extract_zip(source_path, temp_path)
            
        ET.register_namespace('','http://schemas.openxmlformats.org/spreadsheetml/2006/main')
        tree = ET.parse(os.sep.join([temp_path, 'xl', 'sharedStrings.xml']))
        root = tree.getroot()
        for si in root:
            for t in si:
                text = t.text
                local_to_replace = compute_match(t.text, {}, replacements, source_path, update_external)
                for match, value in local_to_replace.items():
                    text = text.replace(match, value)
                t.text = text
        tree.write(os.path.join(temp_path, 'xl', 'sharedStrings.xml'))
        make_zip(temp_path, target_path)
        shutil.rmtree(temp_path)

Helper class that provides a standard way to create an ABC using inheritance.

# Parser()

# def parse(self, path, container, parse_entry, find_matches):

View Source

    def parse(self, path, container, parse_entry, find_matches):
        temp_path = path + '_temp'
        extract_zip(path, temp_path)
        ET.register_namespace('','http://schemas.openxmlformats.org/spreadsheetml/2006/main')
        tree = ET.parse(os.sep.join([temp_path, 'xl', 'sharedStrings.xml']))
        root = tree.getroot()
        for si in root:
            for t in si:
                text = t.text
                matches = find_matches(str(text))
                for match in matches:
                    payload = parse_entry(match, path)
                    container[payload['id']] = payload ## TODO: add check here for duplicates
        shutil.rmtree(temp_path)

Parse file accessible via path property

A general implementation of parse method should include following steps:

Open file
Read file data and transform it's meaningful content into string or an iterable of strings
Call of find_matches function on such strings, resulting in an array of matches
For each match found one should call parse_entry, resulting in an entry dictionary
For each entry use entry "id" property as key and payload as value to populate the container provided

Parameters

path (str): Absolute path to file to be parsed
container (Dict[str, [Dict[str, Any]]]): Dictionary to be populated with parsed entries, contains key-value pairs with entry id property as key and entry payload dictionary as value
parse_entry (callable): Function that extracts entry (current implementation uses json parse) from matching string, returns entry payload dictionary
find_matches (callable): Function that searches the entry string for matches (that is, {{VALID_JSON_OBJECT}} patterns), returns array of matching substrings

# def replace( self, source_path, target_path, compute_match, replacements, update_external=False ):

View Source

    def replace(self, source_path, target_path, compute_match, replacements, update_external = False):
        temp_path = source_path + '_temp'
        try:
            shutil.rmtree(temp_path)
        except Exception as e:
            pass
        extract_zip(source_path, temp_path)
            
        ET.register_namespace('','http://schemas.openxmlformats.org/spreadsheetml/2006/main')
        tree = ET.parse(os.sep.join([temp_path, 'xl', 'sharedStrings.xml']))
        root = tree.getroot()
        for si in root:
            for t in si:
                text = t.text
                local_to_replace = compute_match(t.text, {}, replacements, source_path, update_external)
                for match, value in local_to_replace.items():
                    text = text.replace(match, value)
                t.text = text
        tree.write(os.path.join(temp_path, 'xl', 'sharedStrings.xml'))
        make_zip(temp_path, target_path)
        shutil.rmtree(temp_path)

Replace file contents

A general implementation of replace method should include following steps:

Open file accessible via source_path in read mode
Read file data and transform it so its text content becomes available for editing
For each string in obtained text call compute_match, it results in an dictionary with {{VALID_JSON_OBJECT}} patterns as keys and computed substitutions as values
For each match, value in the dictionary, replace match with value in text
Create file at target_path and write modified text

Parameters

source_path (str): Absolute path to file to be parsed
target_path (str): Absolute path to file to be generated
compute_match (callable): Function that
1. searches the entry string for matches (that is, {{VALID_JSON_OBJECT}} patterns)
2. finds entry id in replacements dictionary
3. populates to_replace dictionary parameter with "{{VALID_JSON_OBJECT}}" as key and replacement string as value
4. if update_external is True, it updates external resources with replacement string
5. returns to_replace dictionary
replacements (Dict[str, str]): Dictionary containing pairs of "{{VALID_JSON_OBJECT}}" keys and their replacements as values
update_external (bool, optional): Boolean, indicating whether external resources should be updated