diff options
| -rw-r--r-- | scrape_ecourtindia_v6/modules/scraper_orders.py | 4 | ||||
| -rw-r--r-- | scrape_ecourtindia_v6/orders_scrape_courts.py | 45 | ||||
| -rw-r--r-- | scrape_ecourtindia_v6/scrape_orders.py | 3 |
3 files changed, 7 insertions, 45 deletions
diff --git a/scrape_ecourtindia_v6/modules/scraper_orders.py b/scrape_ecourtindia_v6/modules/scraper_orders.py index a9fe7be..d0b8df3 100644 --- a/scrape_ecourtindia_v6/modules/scraper_orders.py +++ b/scrape_ecourtindia_v6/modules/scraper_orders.py @@ -71,7 +71,7 @@ class ScraperOrders(Scraper): self.rows.append([ rows[i], rows[i-1].text, rows[i-2].text, rows[i-3].text ]) i += 5 - def handle_orders(self): + def handle_orders(self, court_name): for row in self.rows: order = row[0] @@ -97,7 +97,7 @@ class ScraperOrders(Scraper): except: print(f'UNABLE TO FETCH PDF: {pdf_url}') - record = { 'case_info': row[3], 'petitioner_respondent': row[2], 'date': row[1], 'filename': filename } + record = { 'court_name': court_name, 'case_info': row[3], 'petitioner_respondent': row[2], 'date': row[1], 'filename': filename } self.db.insert(record) self.driver.find_element(By.ID, 'modalOders').find_element(By.CLASS_NAME, 'btn-close').click() diff --git a/scrape_ecourtindia_v6/orders_scrape_courts.py b/scrape_ecourtindia_v6/orders_scrape_courts.py index 597ce9f..d1a4411 100644 --- a/scrape_ecourtindia_v6/orders_scrape_courts.py +++ b/scrape_ecourtindia_v6/orders_scrape_courts.py @@ -22,7 +22,7 @@ class ThreadSafeCSVWriter: def scrape_district(state, district, csv_writer): try: config = {} - scraper = ScraperOrders(config) + scraper = ScraperOrders(None, config) scraper.close_modal() scraper.select('sess_state_code', state) @@ -63,7 +63,7 @@ def scrape_courts(): state = 'Uttar Pradesh' config = {} - scraper = ScraperOrders(config) + scraper = ScraperOrders(None, config) scraper.close_modal() scraper.select('sess_state_code', state) @@ -87,44 +87,5 @@ def scrape_courts(): csv_writer.close() -def scrape_orders(courts): - csvfile = open(courts, newline='') - reader = csv.reader(csvfile) - - for row in reader: - print(row) - config = {} - scraper = ScraperOrders(config) - scraper.close_modal() - - scraper.select('sess_state_code', row[0]) - scraper.select('sess_dist_code', row[1]) - - while True: - sleep(0.5) - try: - modal_is_open = scraper.driver.find_element(By.CLASS_NAME, 'modal').is_displayed() - if modal_is_open: - scraper.close_modal() - continue - break - except: - break - - scraper.select('court_complex_code', row[2]) - sleep(1) - scraper.goto_courtnumber() - - scraper.select('nnjudgecode1', row[3]) - scraper.driver.find_element(By.ID, 'radBoth2').click() - scraper.submit_search() - - scraper.parse_orders_table() - scraper.handle_orders() - - break - - csvfile.close() - if __name__ == '__main__': - scrape_orders('csv/2023-24_pocso.csv') + scrape_courts() diff --git a/scrape_ecourtindia_v6/scrape_orders.py b/scrape_ecourtindia_v6/scrape_orders.py index 54a2d80..b0dc311 100644 --- a/scrape_ecourtindia_v6/scrape_orders.py +++ b/scrape_ecourtindia_v6/scrape_orders.py @@ -43,13 +43,14 @@ def scrape_single_court(row): scraper.select('court_complex_code', row[2]) sleep(1) scraper.goto_courtnumber() + sleep(0.6) scraper.select('nnjudgecode1', row[3]) scraper.driver.find_element(By.ID, 'radBoth2').click() scraper.submit_search() scraper.parse_orders_table() - scraper.handle_orders() + scraper.handle_orders(row[3]) scraper.driver.quit() |