What's Changed

• v6.0.0 Migrate to pyreqwest_impersonate, AsyncDDGS: all functions with a by @deedy5 in https://github.com/deedy5/duckduckgo_search/pull/216

Full Changelog: https://github.com/deedy5/duckduckgo_search/compare/v5.3.1...v6.0.0

• Update requirements
• Mypy: add # type ignore to bypass errors

Full Changelog: https://github.com/deedy5/duckduckgo_search/compare/v5.3.0...v5.3.1

Version with httpx

• Update headers list
• Update requirements

Full Changelog: https://github.com/deedy5/duckduckgo_search/compare/v5.3.0b4...v5.3.1b1

v5.3.0

• add proxy, deprecate proxies parameter
• handling exceptions more gracefully when using asyncio.gather()
• AsyncDDGS|DDGS.del(): improve logic and add check - if hasattr(se…
• AsyncDDGS.del(): suppress RuntimeWarning
• RatelimitException: add Ratelimit to message
• make executor and parser property
• bypass curl_cffi warning on windows

Full Changelog: https://github.com/deedy5/duckduckgo_search/compare/v5.2.2...v5.3.0

v5.3.0b4

• Handling exceptions more gracefully when using asyncio.gather()
• AsyncDDGS.del(): suppress RuntimeWarning

Full Changelog: https://github.com/deedy5/duckduckgo_search/compare/v5.3.0b3...v5.3.0b4

v5.3.0b3

• AsyncDDGS, DDGS: bugfix instance closing

Full Changelog: https://github.com/deedy5/duckduckgo_search/compare/v5.3.0b2...v5.3.0b3

v5.3.0b2

• Migrate from curl_cffi to httpx
• Make orjson optional

Full Changelog: https://github.com/deedy5/duckduckgo_search/compare/v5.2.2...v5.3.0b2

This is a special release that uses the httpx library

• [BUGFIX] Improved instance closing, set stream to False when making a request
• Delay between tests set to 0.5 second
• Update requirements
• Update github workflows

Full Changelog: https://github.com/deedy5/duckduckgo_search/compare/v5.2.1...v5.2.2

• [BUGFIX] AsyncDDGS._text_api(): change payload to prevent blocking

Full Changelog: https://github.com/deedy5/duckduckgo_search/compare/v5.2.0...v5.2.1

What's Changed

• Improve exception handling and add RatelimitException and TimeoutException by @deedy5 in https://github.com/deedy5/duckduckgo_search/pull/201
• Optimize package size: make lxml optional and orjson mandatory by @deedy5 in https://github.com/deedy5/duckduckgo_search/pull/202
• AsyncDDGS._get_executor(): bugfix - using the _executor class variable as a cache
• Extend DDGS and AsyncDDGS for both context-manager and standalone use
• CLI: catch exceptions
• update Readme

Full Changelog: https://github.com/deedy5/duckduckgo_search/compare/v5.1.0...v5.2.0

What's Changed

• BUGFIX: _text_api() - update request payload (api blocks requests)
• Add mypy type-checking, recode to comply by @deedy5 in https://github.com/deedy5/duckduckgo_search/pull/193
• maps(): raise exception if coordinates are not found
• DDGS: make _loop class variable
• AsyncDDGS: make _executor class variable
• utils.py: _is_500_in_url() - decorate with lru_cache
• AsyncDDGS.init(): add 'self._parser' for reuse lxml HtmlParser within class instance

Full Changelog: https://github.com/deedy5/duckduckgo_search/compare/v5.0...v5.1.0

What's Changed

1. [BREAKING CHANGES] DDGS, AsyncDDGS : all functions are now not generators, but return a list of dictionaries
2. Results return is dramatically sped up by asynchronous requests under the hood, especially for large max_results values
3. DDGS: reuse shared event loop,running in thread
4. AsyncDDGS: add global _SHARED_EXECUTOR for using in run_in_executor
5. maps(): removed MapsResult model, results are saved directly by the dictionary
6. maps(): divide bbox only if distance between coordinates > 1km
7. _extract_vqd(): replace regex pattern with .index
8. text(): simplify xpath patterns in _text_html() and _text_lite()
9. using orjson if installed
10. update CLI and tests to keep up with the changes
11. update dependencies: curl-cffi to 0.6.2, ruff to 0.3.0

Full Changelog: https://github.com/deedy5/duckduckgo_search/compare/v4.5.0...v5.0

1. AsyncDDGS, DDGS: remove max_clients parameter (unclear behavior).
2. AsyncDDGS, DDGS: add concurrency parameter to limit the number of concurrent requests (using asyncio.Semaphore).
3. AsyncDDGS.maps(): add squares into queue only if distance between coordinates > 1km.
4. AsyncDDGS.maps(): bugfix exception when work_bboxes_results element is a dict.
5. _extract_vqd(): replace regex pattern with .index().
6. Update dependencies: curl-cffi to 0.6.2. ruff to 0.3.0.

Full Changelog: https://github.com/deedy5/duckduckgo_search/compare/v5.0b...v5.0b1

What's Changed

1. [BRAKING CHANGES!!!] DDGS, AsyncDDGS : all functions are now not generators, but return results as a list of dictionaries,
2. DDGS, AsyncDDGS: added max_clients parameter to regulate the number of concurrent HTTP clients.
3. DDGS, AsyncDDGS: added possibility to use without context manager (it is not necessary to use with block).
4. Results return is dramatically sped up by asynchronous requests under the hood, especially for large max_results values.
5. maps(): removed MapsResult model, results are saved directly by the dictionary.
6. updated CLI and tests to keep up with the changes.
7. updated README.

Full Changelog: https://github.com/deedy5/duckduckgo_search/compare/v4.5.0...v5.0b

What's Changed

1. remove nest-asyncio, run new event loop in thread to bypass 'loop already running',
2. DDGS._iter_over_async() - recode to use queue,
3. AsyncDDGS.translate() - recode to AsyncGenerator, DDGS.translate() - to Generator,
4. update requirements,
5. update README.

v4.5.0 by @deedy5 in https://github.com/deedy5/duckduckgo_search/pull/189

Full Changelog: https://github.com/deedy5/duckduckgo_search/compare/v4.4.3...v4.5.0

1. Bugfix in maps(): no result was returned if only the zip parameter was used

Full Changelog: https://github.com/deedy5/duckduckgo_search/compare/v4.4.2...v4.4.3

What's Changed

• DDGS: using asyncio.run instead of manipulating the loops by @deedy5 in https://github.com/deedy5/duckduckgo_search/pull/186

Full Changelog: https://github.com/deedy5/duckduckgo_search/compare/v4.4.1...v4.4.2

What's Changed

1. DDGS: recode using asyncio loop
2. update curl-cffi requirement version to 0.6.0b9
3. simplify self.proxies logic
4. remove docstring_inheritance from requirements

Full Changelog: https://github.com/deedy5/duckduckgo_search/compare/v4.4...v4.4.1

What's Changed

• Bugfix: add nest_asyncio to run DDGS in an async loop by @deedy5 in https://github.com/deedy5/duckduckgo_search/pull/182

Full Changelog: https://github.com/deedy5/duckduckgo_search/compare/v4.3.2...v4.4

What's Changed

• CLI bugfix: add colors to print maps result by @deedy5 in https://github.com/deedy5/duckduckgo_search/pull/179

Full Changelog: https://github.com/deedy5/duckduckgo_search/compare/v4.3...v4.3.1

What's Changed

• DDGS - remove code, wrap AsyncDDGS by @deedy5 in https://github.com/deedy5/duckduckgo_search/pull/167
• curl-cffi impersonate - always using the latest chrome browser version by @deedy5 in https://github.com/deedy5/duckduckgo_search/pull/176
• Remove async tests (it's enough to test DDGS, because it's inherited from AsyncDDGS) by @deedy5 in https://github.com/deedy5/duckduckgo_search/pull/177
• update MapsResult fields by @deedy5 in https://github.com/deedy5/duckduckgo_search/pull/178

Full Changelog: https://github.com/deedy5/duckduckgo_search/compare/v4.2...v4.3

1. AsyncDDGS performance improved: await resp.acontent(),
2. _get_vqd improved: using regex pattern to extract vqd from html,
3. Removed deprecated functions: ddg, ddg_answers, ddg_images, ddg_maps, ddg_news, ddg_suggestions, ddg_translate, ddg_videos,
4. README: updated async example,
5. Ruff: added lint rules to pyproject.toml,
6. CLI: support functions have been changed to private,
7. Docstrings updated,
8. Small code improvements.

1. _get_url(): bugfix exception message,
2. update README (proxies, bypass curl-cffi NotImplementedError in windows),
3. update bug_report.md

1. random browser in each class instance (headers are set automatically),
2. CLI: file download debugging (removed chunks, added timeout),
3. removed sleep between requests,
4. removed VQDExtractionException,
5. BUGFIX: AsyncDDGS - used AsyncSession instead of Session,
6. BUGFIX AsyncDDGS on windows - change asyncio.set_event_loop_policy
7. logging: improved debug messages.

1. migrate from httpx to curl_cffi
2. CURL: simplified downloads, use ThreadPoolExecutor instead of asyncio
3. exceptions: simplified, use DuckDuckGoSearchException and VQDExtractionException
4. github workflow: added windows and macos
5. tests: removed delays between tests, activated tests for text(backend="lite")
6. CLI: save_json() - set encoding="utf-8" (bugfix for windows)

1. bugfix in images(): httpx.Client(verify=False) to get around an unexpected error in httpx

1. text(backend="api"): extract json from html,
2. bugfix text(backend="html"): update Referer in headers,
3. removed tests for text(backend="lite") - works only with proxies.

1. text(backend="html"/"lite"): removed ad results,
2. README: added exceptions, removed unnecessary async examples.

1. add core exceptions,
2. remove retries in _get_url(),
3. raise ApiException if resp.status_code==403,
4. raise RateLimitException if resp.status_code==202,
5. sleep(0.75) between API requests if proxies is None,
6. sort imports, lint and format .py files with Ruff,
7. pytest: sleep(1) between tests,
8. add "Accept-Encoding" to HEADERS,

1. BUGFIX text(): skip resp if resp.status_code==202,
2. BUGFIX images(): add the "Accept-Language" parameter to the headers.

1. BUGFIX in images() function: add "Accept" parameter to requests,
2. update requirements to use httpx 0.25.1 version (Add support for Python 3.12).

1. BUGFIX - fixing the region parameter in the text() function.
2. BUGFIX - stop redundant requests.

1. update USERAGENTS,
2. update requirements,
3. small improvements to the code,
4. drop python 3.7, add python 3.12.

1. update USERAGENTS,
2. def _text_api(): update payload,
3. update README.md

1. CLI: download_file - bugfix exceptions,
2. CLI: download_file() - change timeout=10,
3. CLI: _download_results() - change default threads=10,
4. CLI - add proxy parameter to news, videos, answers, suggestions, maps, translate.

1. bugfix in DDGS.text(backend='lite'): removed duplicated results
2. CLI: add proxy and threads parameters to ddgs text and ddgs images

1. del sleep() in text(backend='api') and images()
2. CLI: add aiofiles to save files asynchronously
3. _text_api() - del repeated self._normalize(body)
4. _normalize_url(): return '' if not url
5. CLI: small formatting changes
6. update type hints

1. normalize urls (unquote and replace spaces with +),
2. CLI: bugfix in maps() 'source' printing (was splitted),
3. CLI: download files asynchronously.

1. update USERAGENTS (only the most popular ones are left),
2. take a random user-agent when initializing DDGS,
3. bug fix in text(backend="api") - body was not normalized ,
4. bug fix in text(backend="lite") - url was not added to cache to check uniqueness of results,
5. update README.

1. add brotli dependency,
2. update USERAGENTS.

1. translate() - migrate to httpx
2. correct usage socks5 proxies
3. random useragent choice
4. text(backend="lite") - normalize title and body
5. update tests
6. cli: create file only after sucsessful response

Migrate from requests to httpx

1. DDGS().text() - add new parameter backend: api - collect data from duckduckgo.com, html - collect data from html.duckduckgo.com, lite - collect data lite.duckduckgo.com.
2. DDGS().text(backend='api') and DDGS().images() - add sleep(0.75) after receiving the vqd.
3. CLI - ddgs text - add -b (--backend) parameter.
4. add tests for DDGS().text(backend='html') and DDGS().text(backend='lite').

DDGS().text() - bugfix in next_page finding

1. get_url() will raise an exception after unsuccessful retries instead of returning None,
2. assert vqd istead of returning none,
3. update requirements,
4. update README.

Bugfix in useragent: remove anti_useragent

1. Bugfix in _resp_to_json()
2. Sleep 1 sec if HTTPError to prevent ip blocking

Remove '=' from f-strings for python 3.7 compatibility

1. BUGFIX in _get_vqd(). Replace ' by " https://github.com/deedy5/duckduckgo_search/commit/56908f82c7ea5b079ade64842c59f60339afef5e
2. improve logging

refresh_vqd if '506.js' in resp.url

1. don't close VQD_CACHE after each access operation

store the cache in the module directory by default

1. save vqd's to vqd_cache file,
2. add diskcache to requirements.

Folder name error when using site: