solocla/theloftstore
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

extrac prices pdfco

Browse Source
This commit is contained in:
solocla
2025-11-17 15:05:10 +01:00
parent 6b65b31932
commit 1804605cad
16 changed files with 1131 additions and 330 deletions
Download Patch File Download Diff File Expand all files Collapse all files
public/userarea/extract_prices_pdfco.php
+218
View File
@@ -0,0 +1,218 @@
<?php
header("Content-Type: application/json; charset=utf-8");
// ---------------------------------------------------------
// 1) CONFIGURAZIONE
// ---------------------------------------------------------
$apiKey = "info@claudiosironi.com_Qfh02D7sAvi2tcx3ZchHpusNaBquCKhJw81fEnkHe2ersQDVOex4IokhCCzaFAz1";
$fileToken = "filetoken://37bb07a8561409d281b32bcd023ff3dc33de92a8f65aae5af7";
// opzionale: passare un token via GET
if (isset($_GET['token']) && $_GET['token'] !== "") {
$fileToken = $_GET['token'];
}
// ---------------------------------------------------------
// 2) PDF.CO → JSON2
// ---------------------------------------------------------
$endpoint = "https://api.pdf.co/v1/pdf/convert/to/json2";
$payload = [
"url" => $fileToken,
"inline" => true,
"detectTables" => true,
"pages" => ""
];
$ch = curl_init($endpoint);
curl_setopt($ch, CURLOPT_POST, true);
curl_setopt($ch, CURLOPT_HTTPHEADER, [
"Content-Type: application/json",
"x-api-key: $apiKey"
]);
curl_setopt($ch, CURLOPT_POSTFIELDS, json_encode($payload));
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
$response = curl_exec($ch);
if (!$response) {
echo json_encode(["error" => "Errore CURL: " . curl_error($ch)]);
curl_close($ch);
exit;
}
curl_close($ch);
$json = json_decode($response, true);
if (!isset($json["body"]["document"]["page"]["row"])) {
echo json_encode([
"error" => true,
"message" => "JSON PDF.co non contiene tabelle utili",
"raw" => $json
]);
exit;
}
$rows = $json["body"]["document"]["page"]["row"];
// ---------------------------------------------------------
// 3) FUNZIONI UTILITARIE
// ---------------------------------------------------------
// Formato A – 291 02F
function isCodeTypeA($txt)
{
return preg_match('/^\d{3}\s+[0-9A-Z]{2,3}$/', $txt);
}
// Estrai blocchi tipo "499 3A 14 5.475"
function extractTypeCBlocks($line)
{
preg_match_all(
'/(\d{3})\s+([0-9A-Z]{1,3})\s+(\d{2})\s+(\d{1,2}\.\d{3})/',
$line,
$m,
PREG_SET_ORDER
);
$out = [];
foreach ($m as $b) {
$out[] = [
"codice" => $b[1],
"variante" => $b[2],
"dimensione" => $b[3],
"prezzo" => floatval(str_replace(".", "", $b[4]))
];
}
return $out;
}
// ---------------------------------------------------------
// 4) PARSER UNICO (GESTISCE TUTTI I FORMATI)
// ---------------------------------------------------------
$items = [];
$currentCode = null;
$currentDesc = "";
$currentPrices = [];
foreach ($rows as $row) {
// Ricostruisco la riga
$line = "";
foreach ($row["column"] as $col) {
if (!isset($col["text"]["text"])) continue;
$t = trim($col["text"]["text"]);
if ($t !== "") $line .= " " . $t;
}
$line = trim($line);
if ($line === "") continue;
// ---------------------------------------------------------
// 4A — FORMATO CASSINA A (FIX prezzi concatenati)
// ---------------------------------------------------------
if (preg_match('/^(\d{3}\s+[0-9A-Z]{2,3})\s+(.*)$/', $line, $mA)) {
$codice = trim($mA[1]);
$resto = trim($mA[2]);
// Estrae TUTTI i prezzi concatenati (es. "10,808.9906.460...")
preg_match_all('/\d{1,2}[.,]\d{2,3}/', $resto, $matchesPrezzi);
if (count($matchesPrezzi[0]) >= 2) {
// Descrizione SENZA prezzi
$descr = trim(preg_replace('/\d{1,2}[.,]\d{2,3}/', '', $resto));
// Conversione valori
$vals = array_map(function ($v) {
return floatval(str_replace(",", ".", str_replace(".", "", $v)));
}, $matchesPrezzi[0]);
$items[] = [
"type" => "A",
"codice" => $codice,
"descrizione" => $descr,
"prezzi" => $vals
];
continue;
}
}
// ---------------------------------------------------------
// 4B — FORMATO VECCHIO CASSINA (003 BC cromata … 8 prezzi)
// ---------------------------------------------------------
if (preg_match(
'/^(\d{3}\s+[A-Z]{1,3})\s+([A-Za-zÀ-ù\s]+?)\s+((?:\d{1,2}[.,]\d{2,3}\s*){8,})$/',
$line,
$mB
)) {
$codice = trim($mB[1]);
$descr = trim($mB[2]);
$valuesString = trim($mB[3]);
preg_match_all('/\d{1,2}[.,]\d{2,3}/', $valuesString, $vv);
$vals = array_map(function ($v) {
return floatval(str_replace(",", ".", str_replace(".", "", $v)));
}, $vv[0]);
$categorie = ["Z", "Y", "X", "O", "L", "F/COL", "E/COM", "ALTRO"];
$prezziMappati = [];
foreach ($vals as $i => $v) {
$key = $categorie[$i] ?? ("COL_" . ($i + 1));
$prezziMappati[$key] = $v;
}
$items[] = [
"type" => "B1",
"codice" => $codice,
"descrizione" => $descr,
"prezzi" => $prezziMappati
];
continue;
}
// ---------------------------------------------------------
// 4C — FORMATO LISTINO2 (499 3A 14 5.475 ripetuti)
// ---------------------------------------------------------
$blocks = extractTypeCBlocks($line);
if (!empty($blocks)) {
$desc = $line;
foreach ($blocks as $b) {
$pattern = sprintf(
'/%s\s+%s\s+%s\s+\d{1,2}\.\d{3}/',
$b["codice"],
$b["variante"],
$b["dimensione"]
);
$desc = preg_replace($pattern, "", $desc);
}
$desc = trim($desc);
$items[] = [
"type" => "C",
"descrizione" => $desc,
"varianti" => $blocks
];
continue;
}
}
// ---------------------------------------------------------
// 5) OUTPUT FINALE
// ---------------------------------------------------------
echo json_encode([
"status" => "ok",
"total" => count($items),
"items" => $items
], JSON_PRETTY_PRINT);