Skip to content

feat: add get_cc_select_html #496

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
drunkpig merged 1 commit into from
Jul 22, 2025
Merged

feat: add get_cc_select_html #496

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
47 changes: 4 additions & 43 deletions jupyter/main-html-dedup/cc_dedup_fir.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,15 +4,7 @@
"cell_type": "code",
"execution_count": null,
"id": "0",
"metadata": {
"execution": {
"iopub.execute_input": "2025-07-11T02:24:06.149493Z",
"iopub.status.busy": "2025-07-11T02:24:06.149207Z",
"iopub.status.idle": "2025-07-11T02:24:14.378413Z",
"shell.execute_reply": "2025-07-11T02:24:14.377755Z",
"shell.execute_reply.started": "2025-07-11T02:24:06.149476Z"
}
},
"metadata": {},
"outputs": [],
"source": [
"from pyspark.sql import Row\n",
Expand Down Expand Up @@ -57,15 +49,7 @@
"cell_type": "code",
"execution_count": null,
"id": "2",
"metadata": {
"execution": {
"iopub.execute_input": "2025-07-11T02:24:14.379728Z",
"iopub.status.busy": "2025-07-11T02:24:14.379322Z",
"iopub.status.idle": "2025-07-11T02:24:14.382394Z",
"shell.execute_reply": "2025-07-11T02:24:14.381923Z",
"shell.execute_reply.started": "2025-07-11T02:24:14.379710Z"
}
},
"metadata": {},
"outputs": [],
"source": [
"# 获取 cc warc path list\n",
Expand All @@ -81,13 +65,6 @@
"id": "3",
"metadata": {
"editable": true,
"execution": {
"iopub.execute_input": "2025-07-11T02:24:14.383374Z",
"iopub.status.busy": "2025-07-11T02:24:14.383061Z",
"iopub.status.idle": "2025-07-11T02:24:14.397134Z",
"shell.execute_reply": "2025-07-11T02:24:14.396661Z",
"shell.execute_reply.started": "2025-07-11T02:24:14.383359Z"
},
"slideshow": {
"slide_type": ""
},
Expand Down Expand Up @@ -185,15 +162,7 @@
"cell_type": "code",
"execution_count": null,
"id": "4",
"metadata": {
"execution": {
"iopub.execute_input": "2025-07-11T02:24:14.397836Z",
"iopub.status.busy": "2025-07-11T02:24:14.397698Z",
"iopub.status.idle": "2025-07-11T02:24:45.430826Z",
"shell.execute_reply": "2025-07-11T02:24:45.429906Z",
"shell.execute_reply.started": "2025-07-11T02:24:14.397823Z"
}
},
"metadata": {},
"outputs": [],
"source": [
"# mapPartitions 对 warc path 并行解析数据\n",
Expand All @@ -217,15 +186,7 @@
"cell_type": "code",
"execution_count": null,
"id": "6",
"metadata": {
"execution": {
"iopub.execute_input": "2025-07-11T02:24:52.435257Z",
"iopub.status.busy": "2025-07-11T02:24:52.434949Z",
"iopub.status.idle": "2025-07-11T02:26:00.233117Z",
"shell.execute_reply": "2025-07-11T02:26:00.232529Z",
"shell.execute_reply.started": "2025-07-11T02:24:52.435240Z"
}
},
"metadata": {},
"outputs": [],
"source": [
"config[\"skip_output_version\"] = True\n",
Expand Down
144 changes: 21 additions & 123 deletions jupyter/main-html-dedup/cc_dedup_sec.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,13 +4,7 @@
"cell_type": "code",
"execution_count": null,
"id": "0",
"metadata": {
"execution": {
"iopub.status.idle": "2025-07-11T02:26:43.748017Z",
"shell.execute_reply": "2025-07-11T02:26:43.747366Z",
"shell.execute_reply.started": "2025-07-11T02:26:34.645411Z"
}
},
"metadata": {},
"outputs": [],
"source": [
"from xinghe.spark import *\n",
Expand Down Expand Up @@ -50,15 +44,7 @@
"cell_type": "code",
"execution_count": null,
"id": "2",
"metadata": {
"execution": {
"iopub.execute_input": "2025-07-11T02:26:43.749268Z",
"iopub.status.busy": "2025-07-11T02:26:43.748929Z",
"iopub.status.idle": "2025-07-11T02:26:43.751756Z",
"shell.execute_reply": "2025-07-11T02:26:43.751304Z",
"shell.execute_reply.started": "2025-07-11T02:26:43.749250Z"
}
},
"metadata": {},
"outputs": [],
"source": [
"#dump_paths = []\n",
Expand All @@ -71,15 +57,7 @@
"cell_type": "code",
"execution_count": null,
"id": "3",
"metadata": {
"execution": {
"iopub.execute_input": "2025-07-11T02:26:43.752632Z",
"iopub.status.busy": "2025-07-11T02:26:43.752346Z",
"iopub.status.idle": "2025-07-11T02:26:43.762655Z",
"shell.execute_reply": "2025-07-11T02:26:43.762229Z",
"shell.execute_reply.started": "2025-07-11T02:26:43.752617Z"
}
},
"metadata": {},
"outputs": [],
"source": [
"#input_df = spark.read.format(\"json\").load(dump_paths)"
Expand All @@ -97,15 +75,7 @@
"cell_type": "code",
"execution_count": null,
"id": "5",
"metadata": {
"execution": {
"iopub.execute_input": "2025-07-11T02:26:43.763518Z",
"iopub.status.busy": "2025-07-11T02:26:43.763240Z",
"iopub.status.idle": "2025-07-11T02:26:43.770017Z",
"shell.execute_reply": "2025-07-11T02:26:43.769479Z",
"shell.execute_reply.started": "2025-07-11T02:26:43.763503Z"
}
},
"metadata": {},
"outputs": [],
"source": [
"#already_exist_id_v_df = read_any_path(spark, already_exist_id_path, config)"
Expand All @@ -115,15 +85,7 @@
"cell_type": "code",
"execution_count": null,
"id": "6",
"metadata": {
"execution": {
"iopub.execute_input": "2025-07-11T02:26:43.772007Z",
"iopub.status.busy": "2025-07-11T02:26:43.771594Z",
"iopub.status.idle": "2025-07-11T02:26:43.776268Z",
"shell.execute_reply": "2025-07-11T02:26:43.775843Z",
"shell.execute_reply.started": "2025-07-11T02:26:43.771991Z"
}
},
"metadata": {},
"outputs": [],
"source": [
"# 定义 Schema\n",
Expand All @@ -147,15 +109,7 @@
"cell_type": "code",
"execution_count": null,
"id": "8",
"metadata": {
"execution": {
"iopub.execute_input": "2025-07-11T02:26:43.776944Z",
"iopub.status.busy": "2025-07-11T02:26:43.776805Z",
"iopub.status.idle": "2025-07-11T02:26:43.784766Z",
"shell.execute_reply": "2025-07-11T02:26:43.784336Z",
"shell.execute_reply.started": "2025-07-11T02:26:43.776931Z"
}
},
"metadata": {},
"outputs": [],
"source": [
"#join_df = input_df.join(already_exist_id_df, on=\"hash_html\", how=\"left\")\n",
Expand All @@ -173,16 +127,8 @@
{
"cell_type": "code",
"execution_count": null,
"id": "da837b3e-5eee-43b1-9c77-836aac1203df",
"metadata": {
"execution": {
"iopub.execute_input": "2025-07-11T02:26:43.785611Z",
"iopub.status.busy": "2025-07-11T02:26:43.785331Z",
"iopub.status.idle": "2025-07-11T02:27:13.606489Z",
"shell.execute_reply": "2025-07-11T02:27:13.605744Z",
"shell.execute_reply.started": "2025-07-11T02:26:43.785596Z"
}
},
"id": "10",
"metadata": {},
"outputs": [],
"source": [
"input_df = read_any_path(spark, \",\".join(base_input_path), config)"
Expand All @@ -191,16 +137,8 @@
{
"cell_type": "code",
"execution_count": null,
"id": "20135ae2-b1d0-43c9-b241-8d23de94f01d",
"metadata": {
"execution": {
"iopub.execute_input": "2025-07-11T02:27:13.608658Z",
"iopub.status.busy": "2025-07-11T02:27:13.607902Z",
"iopub.status.idle": "2025-07-11T02:27:13.612633Z",
"shell.execute_reply": "2025-07-11T02:27:13.611707Z",
"shell.execute_reply.started": "2025-07-11T02:27:13.608633Z"
}
},
"id": "11",
"metadata": {},
"outputs": [],
"source": [
"def json_data(row_iter)->Row:\n",
Expand All @@ -212,16 +150,8 @@
{
"cell_type": "code",
"execution_count": null,
"id": "0c0435c3-bd5e-4571-ab29-90c522023dec",
"metadata": {
"execution": {
"iopub.execute_input": "2025-07-11T02:27:13.614260Z",
"iopub.status.busy": "2025-07-11T02:27:13.613866Z",
"iopub.status.idle": "2025-07-11T02:27:17.465595Z",
"shell.execute_reply": "2025-07-11T02:27:17.464788Z",
"shell.execute_reply.started": "2025-07-11T02:27:13.614237Z"
}
},
"id": "12",
"metadata": {},
"outputs": [],
"source": [
"undedup_id_df = input_df.rdd.mapPartitions(json_data).toDF()"
Expand All @@ -230,16 +160,8 @@
{
"cell_type": "code",
"execution_count": null,
"id": "10",
"metadata": {
"execution": {
"iopub.execute_input": "2025-07-11T02:27:17.466995Z",
"iopub.status.busy": "2025-07-11T02:27:17.466683Z",
"iopub.status.idle": "2025-07-11T02:27:17.526987Z",
"shell.execute_reply": "2025-07-11T02:27:17.526167Z",
"shell.execute_reply.started": "2025-07-11T02:27:17.466977Z"
}
},
"id": "13",
"metadata": {},
"outputs": [],
"source": [
"def deduplicate_partition(partition):\n",
Expand All @@ -251,24 +173,16 @@
{
"cell_type": "code",
"execution_count": null,
"id": "11",
"metadata": {
"execution": {
"iopub.execute_input": "2025-07-11T02:27:17.528440Z",
"iopub.status.busy": "2025-07-11T02:27:17.528025Z",
"iopub.status.idle": "2025-07-11T02:27:17.537970Z",
"shell.execute_reply": "2025-07-11T02:27:17.537362Z",
"shell.execute_reply.started": "2025-07-11T02:27:17.528418Z"
}
},
"id": "14",
"metadata": {},
"outputs": [],
"source": [
"dedup_df = dedup_part_df.dropDuplicates([\"hash_html\"])"
]
},
{
"cell_type": "markdown",
"id": "12",
"id": "15",
"metadata": {},
"source": [
"# 写出s3"
Expand All @@ -277,16 +191,8 @@
{
"cell_type": "code",
"execution_count": null,
"id": "13",
"metadata": {
"execution": {
"iopub.execute_input": "2025-07-11T02:27:17.539280Z",
"iopub.status.busy": "2025-07-11T02:27:17.539112Z",
"iopub.status.idle": "2025-07-11T02:27:17.581417Z",
"shell.execute_reply": "2025-07-11T02:27:17.580614Z",
"shell.execute_reply.started": "2025-07-11T02:27:17.539264Z"
}
},
"id": "16",
"metadata": {},
"outputs": [],
"source": [
"struct_col = struct(dedup_df[\"track_id\"],dedup_df[\"sub_path\"],dedup_df[\"hash_html\"],)\n",
Expand All @@ -296,16 +202,8 @@
{
"cell_type": "code",
"execution_count": null,
"id": "14",
"metadata": {
"execution": {
"iopub.execute_input": "2025-07-11T02:27:17.582756Z",
"iopub.status.busy": "2025-07-11T02:27:17.582439Z",
"iopub.status.idle": "2025-07-11T02:28:54.423589Z",
"shell.execute_reply": "2025-07-11T02:28:54.422850Z",
"shell.execute_reply.started": "2025-07-11T02:27:17.582737Z"
}
},
"id": "17",
"metadata": {},
"outputs": [],
"source": [
"config[\"skip_output_version\"] = True\n",
Expand Down
Loading