This commit is contained in:
512song@sina.com committed 2022-11-06 21:39:02 +08:00
1 parent c781d24ab0
commit 94bfa87942
1 file changed
+189 -56
+189 -56
View File
@@ -10,27 +10,12 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 16, "execution_count": null,
"id": "1144aaac-92a0-4bcf-aba7-096f7dd3ad3b", "id": "1144aaac-92a0-4bcf-aba7-096f7dd3ad3b",
"metadata": { "metadata": {
"execution": {
"iopub.execute_input": "2022-11-03T11:59:47.933493Z",
"iopub.status.busy": "2022-11-03T11:59:47.932972Z",
"iopub.status.idle": "2022-11-03T11:59:49.278241Z",
"shell.execute_reply": "2022-11-03T11:59:49.276741Z",
"shell.execute_reply.started": "2022-11-03T11:59:47.933444Z"
},
"tags": [] "tags": []
}, },
"outputs": [ "outputs": [],
{
"name": "stdout",
"output_type": "stream",
"text": [
"ok\n"
]
}
],
"source": [ "source": [
"import openpyxl\n", "import openpyxl\n",
"import json\n", "import json\n",
@@ -68,27 +53,12 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 17, "execution_count": null,
"id": "a57ebb22-543c-4804-8d44-89984058ec1d", "id": "a57ebb22-543c-4804-8d44-89984058ec1d",
"metadata": { "metadata": {
"execution": {
"iopub.execute_input": "2022-11-03T12:04:25.033604Z",
"iopub.status.busy": "2022-11-03T12:04:25.033082Z",
"iopub.status.idle": "2022-11-03T12:04:25.402569Z",
"shell.execute_reply": "2022-11-03T12:04:25.401118Z",
"shell.execute_reply.started": "2022-11-03T12:04:25.033556Z"
},
"tags": [] "tags": []
}, },
"outputs": [ "outputs": [],
{
"name": "stdout",
"output_type": "stream",
"text": [
"ok\n"
]
}
],
"source": [ "source": [
"import json\n", "import json\n",
"import time\n", "import time\n",
@@ -143,16 +113,9 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 18, "execution_count": null,
"id": "38397ad4-2bb3-4b5e-8ece-9ea11e801890", "id": "38397ad4-2bb3-4b5e-8ece-9ea11e801890",
"metadata": { "metadata": {
"execution": {
"iopub.execute_input": "2022-11-03T12:04:34.987311Z",
"iopub.status.busy": "2022-11-03T12:04:34.986720Z",
"iopub.status.idle": "2022-11-03T12:04:36.000903Z",
"shell.execute_reply": "2022-11-03T12:04:35.999784Z",
"shell.execute_reply.started": "2022-11-03T12:04:34.987262Z"
},
"tags": [] "tags": []
}, },
"outputs": [], "outputs": [],
@@ -245,20 +208,17 @@
" m_date = result[7]\n", " m_date = result[7]\n",
" if m_date> dict1[user]:\n", " if m_date> dict1[user]:\n",
" dict1[user] = m_date\n", " dict1[user] = m_date\n",
"rq = set() \n", "\n",
"for k, v in dict1.items():\n",
" rq.add(str(v).split(' ')[0].replace('-', '', 2))\n",
"# 创建目录\n",
"m_path = 'file/134'\n", "m_path = 'file/134'\n",
"for pn in rq:\n", "\n",
" if not os.path.exists(m_path + '/' + pn):\n", "fls = glob.glob(f'file/new/*.pdf')\n",
" os.mkdir(m_path + '/' + pn)\n",
"fls = glob.glob(f'file/*.pdf')\n",
"for fn in fls:\n", "for fn in fls:\n",
" #old = os.path.basename(fn).split('.')[0].rjust(8,'0') \n", " #old = os.path.basename(fn).split('.')[0].rjust(8,'0') \n",
" old = os.path.basename(fn).split('.')[0]\n", " old = os.path.basename(fn).split('.')[0]\n",
" mrq = str(dict1[old]).split(' ')[0].replace('-', '', 2)\n", " mrq = str(dict1[old]).split(' ')[0].replace('-', '', 2)\n",
" n_name = f'{m_path}/{mrq}/{str(old).rjust(8,\"0\")}_{mrq}.pdf'\n", " if not os.path.exists(m_path + '/new/' + mrq):\n",
" os.mkdir(m_path + '/new/' + mrq)\n",
" n_name = f'{m_path}/new/{mrq}/{str(old).rjust(8,\"0\")}_{mrq}.pdf'\n",
" if not os.path.exists(n_name):\n", " if not os.path.exists(n_name):\n",
" shutil.copyfile(fn,n_name)\n", " shutil.copyfile(fn,n_name)\n",
"print('ok!')" "print('ok!')"
@@ -314,12 +274,185 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": null, "execution_count": 15,
"id": "a6ab0943-b018-4dd6-9176-85a762f9f60c", "id": "a6ab0943-b018-4dd6-9176-85a762f9f60c",
"metadata": { "metadata": {
"execution": {
"iopub.execute_input": "2022-11-06T12:41:01.250512Z",
"iopub.status.busy": "2022-11-06T12:41:01.250217Z",
"iopub.status.idle": "2022-11-06T12:47:50.596165Z",
"shell.execute_reply": "2022-11-06T12:47:50.595493Z",
"shell.execute_reply.started": "2022-11-06T12:41:01.250488Z"
},
"tags": [] "tags": []
}, },
"outputs": [], "outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"file/134/20221101/new/00030841_20221101.pdf转换成功!\n",
"file/134/20221101/new/03442719_20221101.pdf转换成功!\n",
"file/134/20221101/new/01730401_20221101.pdf转换成功!\n",
"file/134/20221101/new/01735113_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733333_20221101.pdf转换成功!\n",
"file/134/20221101/new/03467856_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733150_20221101.pdf转换成功!\n",
"file/134/20221101/new/01735366_20221101.pdf转换成功!\n",
"file/134/20221101/new/01737694_20221101.pdf转换成功!\n",
"file/134/20221101/new/01734253_20221101.pdf转换成功!\n",
"file/134/20221101/new/03417366_20221101.pdf转换成功!\n",
"file/134/20221101/new/03467840_20221101.pdf转换成功!\n",
"file/134/20221101/new/01732867_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733298_20221101.pdf转换成功!\n",
"file/134/20221101/new/03417184_20221101.pdf转换成功!\n",
"file/134/20221101/new/01734429_20221101.pdf转换成功!\n",
"file/134/20221101/new/01732856_20221101.pdf转换成功!\n",
"file/134/20221101/new/03442655_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733693_20221101.pdf转换成功!\n",
"file/134/20221101/new/03362174_20221101.pdf转换成功!\n",
"file/134/20221101/new/01740008_20221101.pdf转换成功!\n",
"file/134/20221101/new/01734030_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733318_20221101.pdf转换成功!\n",
"file/134/20221101/new/01734044_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733506_20221101.pdf转换成功!\n",
"file/134/20221101/new/03468181_20221101.pdf转换成功!\n",
"file/134/20221101/new/01739084_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733181_20221101.pdf转换成功!\n",
"file/134/20221101/new/01734298_20221101.pdf转换成功!\n",
"file/134/20221101/new/01735744_20221101.pdf转换成功!\n",
"file/134/20221101/new/01739978_20221101.pdf转换成功!\n",
"file/134/20221101/new/03467822_20221101.pdf转换成功!\n",
"file/134/20221101/new/01737695_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733033_20221101.pdf转换成功!\n",
"file/134/20221101/new/03442651_20221101.pdf转换成功!\n",
"file/134/20221101/new/03417283_20221101.pdf转换成功!\n",
"file/134/20221101/new/01737227_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733158_20221101.pdf转换成功!\n",
"file/134/20221101/new/03467911_20221101.pdf转换成功!\n",
"file/134/20221101/new/03442657_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733544_20221101.pdf转换成功!\n",
"file/134/20221101/new/03467866_20221101.pdf转换成功!\n",
"file/134/20221101/new/01732785_20221101.pdf转换成功!\n",
"file/134/20221101/new/01734360_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733279_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733404_20221101.pdf转换成功!\n",
"file/134/20221101/new/03467763_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733044_20221101.pdf转换成功!\n",
"file/134/20221101/new/03467776_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733163_20221101.pdf转换成功!\n",
"file/134/20221101/new/03442789_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733325_20221101.pdf转换成功!\n",
"file/134/20221101/new/01732894_20221101.pdf转换成功!\n",
"file/134/20221101/new/01739172_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733223_20221101.pdf转换成功!\n",
"file/134/20221101/new/01739802_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733026_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733381_20221101.pdf转换成功!\n",
"file/134/20221101/new/01732960_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733388_20221101.pdf转换成功!\n",
"file/134/20221101/new/03467931_20221101.pdf转换成功!\n",
"file/134/20221101/new/01737989_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733178_20221101.pdf转换成功!\n",
"file/134/20221101/new/01730300_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733398_20221101.pdf转换成功!\n",
"file/134/20221101/new/03442667_20221101.pdf转换成功!\n",
"file/134/20221101/new/01735304_20221101.pdf转换成功!\n",
"file/134/20221101/new/01737343_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733086_20221101.pdf转换成功!\n",
"file/134/20221101/new/01737934_20221101.pdf转换成功!\n",
"file/134/20221101/new/03442774_20221101.pdf转换成功!\n",
"file/134/20221101/new/03442715_20221101.pdf转换成功!\n",
"file/134/20221101/new/01734312_20221101.pdf转换成功!\n",
"file/134/20221101/new/01734934_20221101.pdf转换成功!\n",
"file/134/20221101/new/03467909_20221101.pdf转换成功!\n",
"file/134/20221101/new/01732845_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733029_20221101.pdf转换成功!\n",
"file/134/20221101/new/03442804_20221101.pdf转换成功!\n",
"file/134/20221101/new/01732839_20221101.pdf转换成功!\n",
"file/134/20221101/new/01731668_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733863_20221101.pdf转换成功!\n",
"file/134/20221101/new/01739095_20221101.pdf转换成功!\n",
"file/134/20221101/new/01732901_20221101.pdf转换成功!\n",
"file/134/20221101/new/01732989_20221101.pdf转换成功!\n",
"file/134/20221101/new/01734179_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733553_20221101.pdf转换成功!\n",
"file/134/20221101/new/01732929_20221101.pdf转换成功!\n",
"file/134/20221101/new/03467954_20221101.pdf转换成功!\n",
"file/134/20221101/new/03467958_20221101.pdf转换成功!\n",
"file/134/20221101/new/01739856_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733457_20221101.pdf转换成功!\n",
"file/134/20221101/new/03468040_20221101.pdf转换成功!\n",
"file/134/20221101/new/03417280_20221101.pdf转换成功!\n",
"file/134/20221101/new/03442669_20221101.pdf转换成功!\n",
"file/134/20221101/new/01738812_20221101.pdf转换成功!\n",
"file/134/20221101/new/01730159_20221101.pdf转换成功!\n",
"file/134/20221101/new/01732885_20221101.pdf转换成功!\n",
"file/134/20221101/new/03467902_20221101.pdf转换成功!\n",
"file/134/20221101/new/03391210_20221101.pdf转换成功!\n",
"file/134/20221101/new/01739088_20221101.pdf转换成功!\n",
"file/134/20221101/new/01732945_20221101.pdf转换成功!\n",
"file/134/20221101/new/01323056_20221101.pdf转换成功!\n",
"file/134/20221101/new/01735421_20221101.pdf转换成功!\n",
"file/134/20221101/new/03391206_20221101.pdf转换成功!\n",
"file/134/20221101/new/01732910_20221101.pdf转换成功!\n",
"file/134/20221101/new/03467771_20221101.pdf转换成功!\n",
"file/134/20221101/new/01732886_20221101.pdf转换成功!\n",
"file/134/20221101/new/03246101_20221101.pdf转换成功!\n",
"file/134/20221101/new/01730473_20221101.pdf转换成功!\n",
"file/134/20221101/new/01730157_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733320_20221101.pdf转换成功!\n",
"file/134/20221101/new/03467826_20221101.pdf转换成功!\n",
"file/134/20221101/new/01734182_20221101.pdf转换成功!\n",
"file/134/20221101/new/03467813_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733500_20221101.pdf转换成功!\n",
"file/134/20221101/new/01734117_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733032_20221101.pdf转换成功!\n",
"file/134/20221101/new/01735111_20221101.pdf转换成功!\n",
"file/134/20221101/new/01731837_20221101.pdf转换成功!\n",
"file/134/20221101/new/01739997_20221101.pdf转换成功!\n",
"file/134/20221101/new/01319106_20221101.pdf转换成功!\n",
"file/134/20221101/new/01312007_20221101.pdf转换成功!\n",
"file/134/20221101/new/03417268_20221101.pdf转换成功!\n",
"file/134/20221101/new/01735056_20221101.pdf转换成功!\n",
"file/134/20221101/new/03467903_20221101.pdf转换成功!\n",
"file/134/20221101/new/01737731_20221101.pdf转换成功!\n",
"file/134/20221101/new/01730908_20221101.pdf转换成功!\n",
"file/134/20221101/new/01738043_20221101.pdf转换成功!\n",
"file/134/20221101/new/03467795_20221101.pdf转换成功!\n",
"file/134/20221101/new/01732812_20221101.pdf转换成功!\n",
"file/134/20221101/new/01730453_20221101.pdf转换成功!\n",
"file/134/20221101/new/03467930_20221101.pdf转换成功!\n",
"file/134/20221101/new/03442708_20221101.pdf转换成功!\n",
"file/134/20221101/new/03442796_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733365_20221101.pdf转换成功!\n",
"file/134/20221101/new/03467817_20221101.pdf转换成功!\n",
"file/134/20221101/new/03417266_20221101.pdf转换成功!\n",
"file/134/20221101/new/01732854_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733284_20221101.pdf转换成功!\n",
"file/134/20221101/new/03442668_20221101.pdf转换成功!\n",
"file/134/20221101/new/01732926_20221101.pdf转换成功!\n",
"file/134/20221101/new/01732869_20221101.pdf转换成功!\n",
"file/134/20221101/new/01732816_20221101.pdf转换成功!\n",
"file/134/20221101/new/01732890_20221101.pdf转换成功!\n",
"file/134/20221101/new/03442709_20221101.pdf转换成功!\n",
"file/134/20221101/new/03467858_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733275_20221101.pdf转换成功!\n",
"file/134/20221101/new/01739109_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733291_20221101.pdf转换成功!\n",
"file/134/20221101/new/01734037_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733324_20221101.pdf转换成功!\n",
"file/134/20221101/new/01736302_20221101.pdf转换成功!\n",
"file/134/20221101/new/03442728_20221101.pdf转换成功!\n",
"file/134/20221101/new/03442769_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733453_20221101.pdf转换成功!\n",
"file/134/20221101/new/01734776_20221101.pdf转换成功!\n",
"file/134/20221101/new/01735676_20221101.pdf转换成功!\n",
"file/134/20221101/new/01733499_20221101.pdf转换成功!\n",
"file/134/20221101/new/01735677_20221101.pdf转换成功!\n"
]
}
],
"source": [ "source": [
"import fitz\n", "import fitz\n",
"from pdf2image import convert_from_path, convert_from_bytes\n", "from pdf2image import convert_from_path, convert_from_bytes\n",
@@ -356,7 +489,7 @@
" covert2pic(zoom)\n", " covert2pic(zoom)\n",
" pic2pdf(obj)\n", " pic2pdf(obj)\n",
" \n", " \n",
"fi_path = 'file/134/20221028/'\n", "fi_path = 'file/134/20221101/'\n",
"fl = glob.glob(f'{fi_path}*.pdf')\n", "fl = glob.glob(f'{fi_path}*.pdf')\n",
"\n", "\n",
"for fn in fl:\n", "for fn in fl:\n",
@@ -412,7 +545,7 @@
], ],
"metadata": { "metadata": {
"kernelspec": { "kernelspec": {
"display_name": "Python 3", "display_name": "Python 3 (ipykernel)",
"language": "python", "language": "python",
"name": "python3" "name": "python3"
}, },
@@ -426,7 +559,7 @@
"name": "python", "name": "python",
"nbconvert_exporter": "python", "nbconvert_exporter": "python",
"pygments_lexer": "ipython3", "pygments_lexer": "ipython3",
"version": "3.8.10" "version": "3.10.6"
} }
}, },
"nbformat": 4, "nbformat": 4,