desync-data 0.1.3

Utility tools for Desync AI users

desync-data

Utility functions for working with PageData objects from the desync_search library. These tools help you clean, filter, deduplicate, extract links, compute text stats, and export structured data from websites crawled via Desync.

---

🚀 Features

• 🔍 Remove boilerplate from page content (prefix/suffix delimiters)
• 📌 Filter pages by URL substring
• 🧹 Remove duplicate pages (based on text_content)
• 🔗 Extract link graphs (internal navigation structure)
• 📊 Compute text statistics (word count, link density, etc.)
• 📤 Export pages to CSV, JSON, or SQLite

---

📦 Installation

pip install desync-data

---

🧪 Quick Example

from desync_search import DesyncClient
from desync_data import *

client = DesyncClient(api_key="your_api_key")
pages = client.crawl("https://example.com", max_depth=2)

pages = remove_duplicate_pages(pages)
pages = remove_boilerplate_prefix(pages, "START")
pages = remove_boilerplate_suffix(pages, "FOOTER")
pages = filter_by_url_substring(pages, "/team")

save_to_csv(pages, "pages.csv")
save_to_json(pages, "pages.json")
save_to_sqlite(pages, "pages.sqlite")

edges = extract_link_graph(pages)
for stat in map(compute_text_stats, pages):
    print(stat)

---

📚 Function Documentation

remove_boilerplate_prefix(pages, delimiter)

Removes everything before and including the first occurrence of the given delimiter in text_content for each page.

• pages: list of PageData-like objects
• delimiter: string to match and cut before
• Returns: cleaned list of modified objects

---

remove_boilerplate_suffix(pages, delimiter)

Removes everything after and including the first occurrence of the delimiter in text_content.

• pages: list of PageData-like objects
• delimiter: string to cut after
• Returns: cleaned list

---

remove_duplicate_pages(pages)

Eliminates duplicates based on exact text_content. Keeps the first appearance.

• Returns: list with duplicates removed

---

filter_by_url_substring(pages, substring)

Keeps only pages whose url contains the given substring.

• substring: required URL fragment to retain
• Returns: filtered list of pages

---

extract_link_graph(pages, include_external_links=False, only_include_crawled_pages=False)

Returns a list of (source_url, destination_url) pairs based on the internal_links field.

• include_external_links: if False, skip links to other domains
• only_include_crawled_pages: if True, limit to known pages only
• Returns: list of directed edges

---

compute_text_stats(page)

Calculates basic content metrics:

• word_count: total words in text_content
• sentence_count: count of ., !, ?
• unique_word_ratio: unique words / total words
• link_density: words inside <a> tags / total words

Returns: dictionary with all metrics
Requires: beautifulsoup4

---

save_to_csv(pages, filepath, mode="w")

Writes selected fields to a .csv file.

• filepath: output path
• mode: "w" for overwrite, "a" for append
• Fields: url, domain, timestamp, text_content, etc.

---

save_to_json(pages, filepath, mode="w")

Saves all pages as a list of dictionaries to a .json file.

• mode: "w" or "a" (appends if file exists and is valid)

---

save_to_sqlite(pages, db_path, table_name="pages", append=True)

Saves all records to a local SQLite database.

• db_path: file path for .sqlite
• append: if False, drops and recreates table
• table_name: optional table name override

---

🔧 Requirements

• Python 3.7+
• desync_search
• beautifulsoup4 (for text stat parsing)

Install with:

pip install desync-data

---

🧑‍💻 Author

Created by Jackson Ballow for fast and scalable web content processing using Desync.

---

📄 License

MIT License