|
@@ -0,0 +1,198 @@
|
|
1
|
+<?php
|
|
2
|
+defined('BASEPATH') or exit('No direct script access allowed');
|
|
3
|
+
|
|
4
|
+/**
|
|
5
|
+ *
|
|
6
|
+ * Controller ParserCompleto
|
|
7
|
+ *
|
|
8
|
+ * This controller for ...
|
|
9
|
+ *
|
|
10
|
+ * @package CodeIgniter
|
|
11
|
+ * @category Controller CI
|
|
12
|
+ * @author Setiawan Jodi <jodisetiawan@fisip-untirta.ac.id>
|
|
13
|
+ * @author Raul Guerrero <r.g.c@me.com>
|
|
14
|
+ * @link https://github.com/setdjod/myci-extension/
|
|
15
|
+ * @param ...
|
|
16
|
+ * @return ...
|
|
17
|
+ *
|
|
18
|
+ */
|
|
19
|
+
|
|
20
|
+class ParserCompleto extends CI_Controller
|
|
21
|
+{
|
|
22
|
+
|
|
23
|
+ //TODO: PARAMETRIZAR BBDD -> SI EXISTE UN PROCESO EN MARCHA, NO INICIAR SIGUIENTE DESCARGA, AVISAR
|
|
24
|
+ protected $rutaDiaria = FCPATH . "DOWNLOADS/DAILY/";
|
|
25
|
+ protected $rutaDiariaFecha = "";
|
|
26
|
+
|
|
27
|
+ protected $rutaMensual = FCPATH . "DOWNLOADS/MONTH/";
|
|
28
|
+ protected $enlaceAtom = "";
|
|
29
|
+ protected $fechaCreacionLog = "";
|
|
30
|
+
|
|
31
|
+ public function __construct()
|
|
32
|
+ {
|
|
33
|
+ parent::__construct();
|
|
34
|
+ $this->load->helper('file');
|
|
35
|
+ $this->load->helper('xml');
|
|
36
|
+ $this->load->library('parserfile');
|
|
37
|
+
|
|
38
|
+ $this->load->model("Parser_model");
|
|
39
|
+ }
|
|
40
|
+
|
|
41
|
+ public function index()
|
|
42
|
+ {
|
|
43
|
+ $existCronRun = $this->Parser_model->existCronRun();
|
|
44
|
+ $fechaCarpeta = "";
|
|
45
|
+ if ($existCronRun['num_rows'] > 0) {
|
|
46
|
+ if ($existCronRun['datos']->cron_ejecucion == 0) {
|
|
47
|
+
|
|
48
|
+ $this->Parser_model->inicioParser();
|
|
49
|
+
|
|
50
|
+ $resultSetPerfiles = $this->Parser_model->getUrlArchivos();
|
|
51
|
+ $fechaCarpeta = date("Y-m-d_H-i-s");
|
|
52
|
+ $fechaImportacionLog = date("Y-m-d H:i:s");
|
|
53
|
+
|
|
54
|
+ foreach ($resultSetPerfiles['datos'] as $row) {
|
|
55
|
+
|
|
56
|
+ $idUltimaImportacion = $this->Parser_model->setFechaNuevaImportacion($fechaImportacionLog);
|
|
57
|
+ $this->fechaCreacionLog = $fechaCarpeta;
|
|
58
|
+ $rutaPerfil = $this->rutaDiaria . "/" . $row->prefijos_tablas;
|
|
59
|
+
|
|
60
|
+ $this->newFolder($rutaPerfil);
|
|
61
|
+
|
|
62
|
+ if ($this->newFolder($rutaPerfil . "/" . $fechaCarpeta)) {
|
|
63
|
+ $this->rutaDiariaFecha = $rutaPerfil . "/" . $fechaCarpeta . "/";
|
|
64
|
+ $this->downloadProcess($row->url_descarga . ".atom");
|
|
65
|
+ }
|
|
66
|
+
|
|
67
|
+ $files = get_filenames($this->rutaDiariaFecha . "/format");
|
|
68
|
+ emailInicioProcesoDescarga($fechaImportacionLog, $files);
|
|
69
|
+
|
|
70
|
+ $this->parserfile->index($this->rutaDiariaFecha, $fechaCarpeta, $row, $idUltimaImportacion);
|
|
71
|
+ }
|
|
72
|
+
|
|
73
|
+ $this->tablasCalculadas($idUltimaImportacion);
|
|
74
|
+
|
|
75
|
+ emailFinProcesoDescarga($fechaImportacionLog);
|
|
76
|
+
|
|
77
|
+ } else {
|
|
78
|
+ echo "EXISTE UN CRON EN MARCHA.";
|
|
79
|
+ }
|
|
80
|
+ } else {
|
|
81
|
+ echo "No existe parametro que iniciar.";
|
|
82
|
+ }
|
|
83
|
+
|
|
84
|
+ $this->Parser_model->stopParser();
|
|
85
|
+
|
|
86
|
+ /* $resultSetPerfiles = $this->Parser_model->getUrlArchivos();
|
|
87
|
+ foreach ($resultSetPerfiles['datos'] as $row) {
|
|
88
|
+ $this->parserfile->index($this->rutaDiaria . "PER_PCSP_EX/2022-02-17_20-38-16", date("Y-m-d"), $row);
|
|
89
|
+ } */
|
|
90
|
+ }
|
|
91
|
+
|
|
92
|
+ private function downloadProcess($urlXML)
|
|
93
|
+ {
|
|
94
|
+
|
|
95
|
+ $fileName = basename($urlXML);
|
|
96
|
+ $this->downloadFile($fileName, $urlXML, $this->rutaDiariaFecha);
|
|
97
|
+ $archivoAtom = $this->getFile($this->rutaDiariaFecha . $fileName, $fileName);
|
|
98
|
+ $nuevoAtomAttrNext = $this->readNextAttribute($archivoAtom);
|
|
99
|
+
|
|
100
|
+ while ($nuevoAtomAttrNext !== "") {
|
|
101
|
+
|
|
102
|
+ $fileName = basename($nuevoAtomAttrNext);
|
|
103
|
+ $this->downloadFile($fileName, $nuevoAtomAttrNext, $this->rutaDiariaFecha);
|
|
104
|
+ $archivoAtom = $this->getFile($this->rutaDiariaFecha . $fileName, $fileName);
|
|
105
|
+
|
|
106
|
+ $nuevoAtomAttrNext = $this->readNextAttribute($archivoAtom);
|
|
107
|
+ }
|
|
108
|
+ }
|
|
109
|
+
|
|
110
|
+ private function readNextAttribute($archivo)
|
|
111
|
+ {
|
|
112
|
+ $enlace = "";
|
|
113
|
+ foreach ($archivo['link'] as $row) {
|
|
114
|
+ $attrNext = $row['@attributes']['rel'];
|
|
115
|
+ $attrHref = $row['@attributes']['href'];
|
|
116
|
+
|
|
117
|
+ if ($attrNext === "next") {
|
|
118
|
+ if (!strpos($attrHref, "licitacionesPerfilesContratanteCompleto3_2021")) { //"20220127_")) {
|
|
119
|
+ $enlace = $attrHref;
|
|
120
|
+ } else {
|
|
121
|
+ $enlace = "";
|
|
122
|
+ }
|
|
123
|
+ //$enlace = $attrHref;
|
|
124
|
+ }
|
|
125
|
+ }
|
|
126
|
+
|
|
127
|
+ return $enlace;
|
|
128
|
+ }
|
|
129
|
+
|
|
130
|
+ private function getFile($url, $fileName)
|
|
131
|
+ {
|
|
132
|
+
|
|
133
|
+ $feed = implode(file($url));
|
|
134
|
+ $xml = simplexml_load_string($feed);
|
|
135
|
+ $json = json_encode($xml);
|
|
136
|
+ $array = json_decode($json, true);
|
|
137
|
+
|
|
138
|
+ $this->formatXml($feed, $fileName);
|
|
139
|
+ return $array;
|
|
140
|
+ }
|
|
141
|
+
|
|
142
|
+ private function newFolder($rutaFinal)
|
|
143
|
+ {
|
|
144
|
+ $result = false;
|
|
145
|
+ if (!is_dir($rutaFinal)) {
|
|
146
|
+ $result = mkdir($rutaFinal);
|
|
147
|
+ mkdir($rutaFinal . "/format");
|
|
148
|
+ }
|
|
149
|
+
|
|
150
|
+ return $result;
|
|
151
|
+ }
|
|
152
|
+
|
|
153
|
+ private function formatXml($xml, $fileName)
|
|
154
|
+ {
|
|
155
|
+ $openingFormat = $this->regExIniciosEsquema($xml);
|
|
156
|
+ $closureFormat = $this->regExFinalesEsquema($openingFormat);
|
|
157
|
+ file_put_contents($this->rutaDiariaFecha . "/format/" . $fileName, $closureFormat);
|
|
158
|
+ }
|
|
159
|
+
|
|
160
|
+ private function downloadFile($fileName, $url, $ruta)
|
|
161
|
+ {
|
|
162
|
+ if (file_put_contents($ruta . $fileName, file_get_contents($url))) {
|
|
163
|
+ return true;
|
|
164
|
+ } else {
|
|
165
|
+ return false;
|
|
166
|
+ }
|
|
167
|
+ }
|
|
168
|
+
|
|
169
|
+ private function regExIniciosEsquema($xml)
|
|
170
|
+ {
|
|
171
|
+ $pattern = "/(<cac:)|(<cbc:)|(<cac-place-ext:)|(<cbc-place-ext:)/";
|
|
172
|
+ $result = preg_replace($pattern, "<", $xml);
|
|
173
|
+
|
|
174
|
+ return $result;
|
|
175
|
+ }
|
|
176
|
+
|
|
177
|
+ private function regExFinalesEsquema($xml)
|
|
178
|
+ {
|
|
179
|
+ $pattern = "/(<\/cac:)|(<\/cbc:)|(<\/cac-place-ext:)|(<\/cbc-place-ext:)/";
|
|
180
|
+ $result = preg_replace($pattern, "</", $xml);
|
|
181
|
+
|
|
182
|
+ return $result;
|
|
183
|
+ }
|
|
184
|
+
|
|
185
|
+ private function tablasCalculadas($idUltimaImportacion)
|
|
186
|
+ {
|
|
187
|
+ $this->Parser_model->truncateTables("vista_datos_licitacion");
|
|
188
|
+
|
|
189
|
+ $licitaciones = $this->Xlsexport_model->getLicitaciones($idUltimaImportacion);
|
|
190
|
+
|
|
191
|
+ insertBucleDuplicateOnKey($licitaciones['datos'], "vista_datos_licitacion", "");
|
|
192
|
+
|
|
193
|
+ }
|
|
194
|
+
|
|
195
|
+}
|
|
196
|
+
|
|
197
|
+/* End of file ParserCompleto.php */
|
|
198
|
+/* Location: ./application/controllers/ParserCompleto.php */
|