1pub mod error;
2pub mod extractors;
3pub mod models;
4pub mod parsers;
5
6pub use error::{InvoiceParserError, Result};
7pub use extractors::{PdfDocument, PdfExtractor, SheetData, XlsxExtractor};
8pub use models::{
9 Address, Currency, DocumentFormat, Invoice, InvoiceType, InvoiceValidation, LineItem,
10 LineItemValidation, ParseResult, Party, PaymentInfo, TaxSummary,
11};
12pub use parsers::InvoiceParser;
13
14use std::path::Path;
15
16pub fn parse_file<P: AsRef<Path>>(path: P) -> Result<ParseResult> {
17 InvoiceParser::parse_file(path)
18}
19
20pub fn parse_pdf<P: AsRef<Path>>(path: P) -> Result<ParseResult> {
21 InvoiceParser::parse_pdf(path)
22}
23
24pub fn parse_pdf_bytes(bytes: &[u8]) -> Result<ParseResult> {
25 InvoiceParser::parse_pdf_bytes(bytes)
26}
27
28pub fn parse_xlsx<P: AsRef<Path>>(path: P) -> Result<ParseResult> {
29 InvoiceParser::parse_xlsx(path)
30}
31
32pub fn parse_text(text: &str) -> Result<Invoice> {
33 InvoiceParser::parse_text(text)
34}
35
36#[cfg(test)]
37mod tests {
38 use super::*;
39 use chrono::Datelike;
40
41 #[test]
42 fn test_parse_simple_invoice_text() {
43 let text = r#"
44 INVOICE
45 Invoice Number: INV-2024-001
46 Invoice Date: 2024-01-15
47 Due Date: 2024-02-15
48
49 Subtotal: $1,000.00
50 Tax: $100.00
51 Total: $1,100.00
52 "#;
53
54 let invoice = parse_text(text).unwrap();
55
56 assert_eq!(invoice.invoice_number, Some("INV-2024-001".to_string()));
57 assert_eq!(invoice.invoice_type, InvoiceType::Standard);
58 assert_eq!(invoice.total_amount, 1100.0);
59 assert_eq!(invoice.subtotal, Some(1000.0));
60 assert_eq!(invoice.total_tax, Some(100.0));
61 }
62
63 #[test]
64 fn test_parse_invoice_with_currency() {
65 let text = "Invoice #12345\nTotal: EUR 500.00";
66 let invoice = parse_text(text).unwrap();
67
68 assert_eq!(invoice.invoice_number, Some("12345".to_string()));
69 assert_eq!(invoice.currency, Currency::EUR);
70 }
71
72 #[test]
73 fn test_detect_credit_note() {
74 let text = "CREDIT NOTE\nReference: CN-001\nTotal: $200.00";
75 let invoice = parse_text(text).unwrap();
76
77 assert_eq!(invoice.invoice_type, InvoiceType::CreditNote);
78 }
79
80 #[test]
81 fn test_detect_receipt() {
82 let text = "RECEIPT\nReceipt #R-001\nAmount: $50.00";
83 let invoice = parse_text(text).unwrap();
84
85 assert_eq!(invoice.invoice_type, InvoiceType::Receipt);
86 }
87
88 #[test]
89 fn test_currency_from_str() {
90 assert_eq!(Currency::from("USD"), Currency::USD);
91 assert_eq!(Currency::from("$"), Currency::USD);
92 assert_eq!(Currency::from("EUR"), Currency::EUR);
93 assert_eq!(Currency::from("€"), Currency::EUR);
94 assert_eq!(Currency::from("GBP"), Currency::GBP);
95 assert_eq!(Currency::from("£"), Currency::GBP);
96 assert_eq!(Currency::from("CNY"), Currency::CNY);
97 assert_eq!(Currency::from("XYZ"), Currency::Other("XYZ".to_string()));
98 }
99
100 #[test]
101 fn test_invoice_to_json() {
102 let text = "Invoice #TEST-001\nTotal: $100.00";
103 let invoice = parse_text(text).unwrap();
104 let json = invoice.to_json().unwrap();
105
106 assert!(json.contains("TEST-001"));
107 assert!(json.contains("100"));
108 }
109
110 #[test]
111 fn test_empty_document_error() {
112 let result = parse_text("");
113 assert!(result.is_err());
114 }
115
116 #[test]
117 fn test_parse_result_with_source() {
118 let invoice = Invoice::new();
119 let result = ParseResult::single(invoice).with_source("test.pdf");
120
121 assert_eq!(result.source_file, Some("test.pdf".to_string()));
122 assert_eq!(result.invoices.len(), 1);
123 }
124
125 #[test]
126 fn test_aws_account_number_extraction() {
127 let text = "Account No：200221655242 (Mlytics Limited-200221655242)\nTotal Amount Due\nUSD 13586.67";
128 let invoice = parse_text(text).unwrap();
129
130 assert_eq!(invoice.invoice_number, Some("200221655242".to_string()));
131 }
132
133 #[test]
134 fn test_account_name_extraction() {
135 let text = "Account No: 970363902599 ( mlytics-prdservice2 )\nCompany name: Mlytics Limited\nTotal: $47.59";
136 let invoice = parse_text(text).unwrap();
137
138 assert_eq!(invoice.invoice_number, Some("970363902599".to_string()));
139 assert_eq!(
140 invoice.account_name,
141 Some("mlytics-prdservice2".to_string())
142 );
143 }
144
145 #[test]
146 fn test_aws_vendor_detection() {
147 let text = "AWS Services Pricing 14,301.76\nTotal: $100.00";
148 let invoice = parse_text(text).unwrap();
149
150 assert_eq!(invoice.vendor.name, Some("Amazon Web Services".to_string()));
151 }
152
153 #[test]
154 fn test_ecloudvalley_vendor_detection() {
155 let text = "伊雲谷(香港) ECLOUDVALLEY TECHNOLOGY (HK) LIMITED\nTotal: $100.00";
156 let invoice = parse_text(text).unwrap();
157
158 assert_eq!(
159 invoice.vendor.name,
160 Some("eCloudValley Technology".to_string())
161 );
162 }
163
164 #[test]
165 fn test_customer_extraction_from_company_name() {
166 let text = "Company name: Mlytics Limited\nBilling Address: Some Address\nTotal: $100.00";
167 let invoice = parse_text(text).unwrap();
168
169 assert_eq!(invoice.customer.name, Some("Mlytics Limited".to_string()));
170 }
171
172 #[test]
173 fn test_aws_service_charge_subtotal() {
174 let text = "AWS Service Charge 1,806.55\nDiscount -108.39\nTotal Amount Due\nUSD 1,698.16";
175 let invoice = parse_text(text).unwrap();
176
177 assert_eq!(invoice.subtotal, Some(1806.55));
178 }
179
180 #[test]
181 fn test_aws_services_pricing_subtotal() {
182 let text = "AWS Services Pricing 14,301.76\nDiscount -715.09\nTotal: $13,586.67";
183 let invoice = parse_text(text).unwrap();
184
185 assert_eq!(invoice.subtotal, Some(14301.76));
186 }
187
188 #[test]
189 fn test_line_items_extraction_from_pdf() {
190 let result = parse_file("examples/pdf/202512-CH306-7133-970363902599.pdf").unwrap();
191 let invoice = &result.invoices[0];
192
193 assert!(!invoice.line_items.is_empty());
194
195 let vpc_item = invoice.line_items.iter().find(|item| {
196 item.service_name
197 .as_ref()
198 .is_some_and(|s| s.contains("Virtual Private Cloud"))
199 });
200 assert!(vpc_item.is_some());
201 assert_eq!(
202 vpc_item.unwrap().description,
203 "USE1-PublicIPv4:InUseAddress"
204 );
205 }
206
207 #[test]
208 fn test_line_items_have_service_name_and_description() {
209 let result = parse_file("examples/pdf/202512-CH306-7133-970363902599.pdf").unwrap();
210 let invoice = &result.invoices[0];
211
212 for item in &invoice.line_items {
213 assert!(item.service_name.is_some());
214 assert!(!item.description.is_empty());
215 }
216 }
217
218 #[test]
219 fn test_line_items_have_unit_price() {
220 let result = parse_file("examples/pdf/HPK_AWS_eCv_200221655242_202512.pdf").unwrap();
221 let invoice = &result.invoices[0];
222
223 let item_with_price = invoice
224 .line_items
225 .iter()
226 .find(|item| item.unit_price.is_some());
227 assert!(item_with_price.is_some());
228 }
229
230 #[test]
231 fn test_statement_type_detection() {
232 let text = "Description of Statement Currency: USD\nTotal: $100.00";
233 let invoice = parse_text(text).unwrap();
234
235 assert_eq!(invoice.invoice_type, InvoiceType::Statement);
236 }
237
238 #[test]
239 fn test_date_format_yyyy_mm_dd() {
240 let text = "Invoice Date: 2025-01-15\nTotal: $100.00";
241 let invoice = parse_text(text).unwrap();
242
243 assert!(invoice.invoice_date.is_some());
244 let date = invoice.invoice_date.unwrap();
245 assert_eq!(date.year(), 2025);
246 assert_eq!(date.month(), 1);
247 assert_eq!(date.day(), 15);
248 }
249
250 #[test]
251 fn test_date_format_yyyy_slash() {
252 let text = "Payment Due Date\n2025/02/25\nTotal: $100.00";
253 let invoice = parse_text(text).unwrap();
254
255 assert!(invoice.due_date.is_some());
256 let date = invoice.due_date.unwrap();
257 assert_eq!(date.year(), 2025);
258 assert_eq!(date.month(), 2);
259 assert_eq!(date.day(), 25);
260 }
261
262 #[test]
263 fn test_chinese_colon_support() {
264 let text = "Account No：123456789\nCompany name：Test Company\nTotal: $100.00";
265 let invoice = parse_text(text).unwrap();
266
267 assert_eq!(invoice.invoice_number, Some("123456789".to_string()));
268 assert_eq!(invoice.customer.name, Some("Test Company".to_string()));
269 }
270
271 #[test]
272 fn test_amount_with_comma_separator() {
273 let text = "Grand Total: $13,586.67";
274 let invoice = parse_text(text).unwrap();
275
276 assert_eq!(invoice.total_amount, 13586.67);
277 }
278}