Duplicate from bigscience-data/roots-search

Co-authored-by: Aleksandra Piktus <[email protected]>

.gitattributes

@@ -0,0 +1,34 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+*.pdf filter=lfs diff=lfs merge=lfs -text

Makefile

@@ -0,0 +1,7 @@
+.PHONY: style
+
+# Format source code automatically
+
+style:
+	black --line-length 119 --target-version py36 .
+	isort .

README.md

@@ -0,0 +1,14 @@
+---
+title: Roots Search Tool
+emoji: 🌸 🔎
+colorFrom: green
+colorTo: indigo
+sdk: gradio
+sdk_version: 3.7
+app_file: app.py
+pinned: false
+license: apache-2.0
+duplicated_from: bigscience-data/roots-search
+---
+
+Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

app.py

@@ -0,0 +1,255 @@
+import http.client as http_client
+import json
+import logging
+import os
+import re
+import string
+
+import gradio as gr
+import requests
+from huggingface_hub import HfApi
+
+hf_api = HfApi()
+roots_datasets = {dset.id.split("/")[-1]:dset for dset in hf_api.list_datasets(author="bigscience-data", use_auth_token=os.environ.get("bigscience_data_token"))}
+
+def get_docid_html(docid):
+    data_org, dataset, docid = docid.split("/")
+    metadata = roots_datasets[dataset]
+    if metadata.private:
+        docid_html = (
+            f"<a "
+            f'class="underline-on-hover"'
+            f'title="This dataset is private. See the introductory text for more information"'
+            f'style="color:#AA4A44;"'
+            f'href="https://huggingface.co/datasets/bigscience-data/{dataset}"'
+            f'target="_blank"><b>🔒{dataset}</b></a><span style="color: #7978FF;">/{docid}</span>'
+        )
+    else:
+        docid_html = (
+            f"<a "
+            f'class="underline-on-hover"'
+            f'title="This dataset is licensed {metadata.tags[0].split(":")[-1]}"'
+            f'style="color:#2D31FA;"'
+            f'href="https://huggingface.co/datasets/bigscience-data/{dataset}"'
+            f'target="_blank"><b>{dataset}</b></a><span style="color: #7978FF;">/{docid}</span>'
+        )        
+    return docid_html
+
+
+PII_TAGS = {"KEY", "EMAIL", "USER", "IP_ADDRESS", "ID", "IPv4", "IPv6"}
+PII_PREFIX = "PI:"
+
+
+def process_pii(text):
+    for tag in PII_TAGS:
+        text = text.replace(
+            PII_PREFIX + tag,
+            """<b><mark style="background: Fuchsia; color: Lime;">REDACTED {}</mark></b>""".format(tag),
+        )
+    return text
+
+
+def process_results(results, highlight_terms):
+    if len(results) == 0:
+        return """<br><p style='font-family: Arial; color:Silver; text-align: center;'>
+                No results retrieved.</p><br><hr>"""
+
+    results_html = ""
+    for result in results:
+        tokens = result["text"].split()
+        tokens_html = []
+        for token in tokens:
+            if token in highlight_terms:
+                tokens_html.append("<b>{}</b>".format(token))
+            else:
+                tokens_html.append(token)
+        tokens_html = " ".join(tokens_html)
+        tokens_html = process_pii(tokens_html)
+        meta_html = (
+            """
+                <p class='underline-on-hover' style='font-size:12px; font-family: Arial; color:#585858; text-align: left;'>
+                <a href='{}' target='_blank'>{}</a></p>""".format(
+                result["meta"]["url"], result["meta"]["url"]
+            )
+            if "meta" in result and result["meta"] is not None and "url" in result["meta"]
+            else ""
+        )
+        docid_html = get_docid_html(result["docid"])
+        results_html += """{}
+            <p style='font-size:14px; font-family: Arial; color:#7978FF; text-align: left;'>Document ID: {}</p>
+            <p style='font-size:12px; font-family: Arial; color:MediumAquaMarine'>Language: {}</p>
+            <p style='font-family: Arial;'>{}</p>
+            <br>
+        """.format(
+            meta_html, docid_html, result["lang"], tokens_html
+        )
+    return results_html + "<hr>"
+
+
+def scisearch(query, language, num_results=10):
+    try:
+        query = " ".join(query.split())
+        if query == "" or query is None:
+            return ""
+
+        post_data = {"query": query, "k": num_results}
+        if language != "detect_language":
+            post_data["lang"] = language
+
+        output = requests.post(
+            os.environ.get("address"),
+            headers={"Content-type": "application/json"},
+            data=json.dumps(post_data),
+            timeout=60,
+        )
+
+        payload = json.loads(output.text)
+
+        if "err" in payload:
+            if payload["err"]["type"] == "unsupported_lang":
+                detected_lang = payload["err"]["meta"]["detected_lang"]
+                return f"""
+                    <p style='font-size:18px; font-family: Arial; color:MediumVioletRed; text-align: center;'>
+                    Detected language <b>{detected_lang}</b> is not supported.<br>
+                    Please choose a language from the dropdown or type another query.
+                    </p><br><hr><br>"""
+
+        results = payload["results"]
+        highlight_terms = payload["highlight_terms"]
+
+        if language == "detect_language":
+            return (
+                (
+                    f"""<p style='font-family: Arial; color:MediumAquaMarine; text-align: center; line-height: 3em'>
+                Detected language: <b>{results[0]["lang"]}</b></p><br><hr><br>"""
+                    if len(results) > 0 and language == "detect_language"
+                    else ""
+                )
+                + process_results(results, highlight_terms)
+            )
+
+        if language == "all":
+            results_html = ""
+            for lang, results_for_lang in results.items():
+                if len(results_for_lang) == 0:
+                    results_html += f"""<p style='font-family: Arial; color:Silver; text-align: left; line-height: 3em'>
+                            No results for language: <b>{lang}</b><hr></p>"""
+                    continue
+
+                collapsible_results = f"""
+                    <details>
+                        <summary style='font-family: Arial; color:MediumAquaMarine; text-align: left; line-height: 3em'>
+                            Results for language: <b>{lang}</b><hr>
+                        </summary>
+                        {process_results(results_for_lang, highlight_terms)}
+                    </details>"""
+                results_html += collapsible_results
+            return results_html
+
+        return process_results(results, highlight_terms)
+
+    except Exception as e:
+        results_html = f"""
+                <p style='font-size:18px; font-family: Arial; color:MediumVioletRed; text-align: center;'>
+                Raised {type(e).__name__}</p>
+                <p style='font-size:14px; font-family: Arial; '>
+                Check if a relevant discussion already exists in the Community tab. If not, please open a discussion.
+                </p>
+            """
+
+    return results_html
+
+
+def flag(query, language, num_results, issue_description):
+    try:
+        post_data = {"query": query, "k": num_results, "flag": True, "description": issue_description}
+        if language != "detect_language":
+            post_data["lang"] = language
+
+        output = requests.post(
+            os.environ.get("address"),
+            headers={"Content-type": "application/json"},
+            data=json.dumps(post_data),
+            timeout=120,
+        )
+
+        results = json.loads(output.text)
+    except:
+        print("Error flagging")
+    return ""
+
+
+description = """# <p style="text-align: center;"> 🌸 🔎 ROOTS search tool 🔍 🌸 </p>
+The ROOTS corpus was developed during the [BigScience workshop](https://bigscience.huggingface.co/) for the purpose
+of training the Multilingual Large Language Model [BLOOM](https://huggingface.co/bigscience/bloom). This tool allows
+you to search through the ROOTS corpus. We serve a BM25 index for each language or group of languages included in
+ROOTS. You can read more about the details of the tool design
+[here](https://huggingface.co/spaces/bigscience-data/scisearch/blob/main/roots_search_tool_specs.pdf). For more
+information and instructions on how to access the full corpus check [this form](https://forms.gle/qyYswbEL5kA23Wu99)."""
+
+
+if __name__ == "__main__":
+    demo = gr.Blocks(
+        css=".underline-on-hover:hover { text-decoration: underline; } .flagging { font-size:12px; color:Silver; }"
+    )
+
+    with demo:
+        with gr.Row():
+            gr.Markdown(value=description)
+        with gr.Row():
+            query = gr.Textbox(lines=1, max_lines=1, placeholder="Type your query here...", label="Query")
+        with gr.Row():
+            lang = gr.Dropdown(
+                choices=[
+                    "ar",
+                    "ca",
+                    "code",
+                    "en",
+                    "es",
+                    "eu",
+                    "fr",
+                    "id",
+                    "indic",
+                    "nigercongo",
+                    "pt",
+                    "vi",
+                    "zh",
+                    "detect_language",
+                    "all",
+                ],
+                value="en",
+                label="Language",
+            )
+        with gr.Row():
+            k = gr.Slider(1, 100, value=10, step=1, label="Max Results")
+        with gr.Row():
+            submit_btn = gr.Button("Submit")
+        with gr.Row():
+            results = gr.HTML(label="Results")
+        flag_description = """
+            <p class='flagging'>
+            If you choose to flag your search, we will save the query, language and the number of results you requested.
+            Please consider adding any additional context in the box on the right.</p>"""
+        with gr.Column(visible=False) as flagging_form:
+            flag_txt = gr.Textbox(
+                lines=1,
+                placeholder="Type here...",
+                label="""If you choose to flag your search, we will save the query, language and the number of results
+                    you requested. Please consider adding relevant additional context below:""",
+            )
+            flag_btn = gr.Button("Flag Results")
+            flag_btn.click(flag, inputs=[query, lang, k, flag_txt], outputs=[flag_txt])
+
+        def submit(query, lang, k):
+            query = query.strip()
+            if query is None or query == "":
+                return "", ""
+            return {
+                results: scisearch(query, lang, k),
+                flagging_form: gr.update(visible=True),
+            }
+
+        query.submit(fn=submit, inputs=[query, lang, k], outputs=[results, flagging_form])
+        submit_btn.click(submit, inputs=[query, lang, k], outputs=[results, flagging_form])
+
+    demo.launch(enable_queue=True, debug=True)

roots_search_tool_specs.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:44cf8caa3819e8abf036178c4d329363d0649def28ec4e0979a0e0b2b94362d9
+size 2642911

spaces.code-workspace

@@ -0,0 +1,8 @@
+{
+	"folders": [
+		{
+			"path": ".."
+		}
+	],
+	"settings": {}
+}