Fixed companies / company pages from name to domain

ddxv · Oct 14, 2024 · 522ed9e · 522ed9e
1 parent 47780c7
commit 522ed9e
Show file tree

Hide file tree

Showing 5 changed files with 62 additions and 35 deletions.
diff --git a/backend/api_app/controllers/companies.py b/backend/api_app/controllers/companies.py
@@ -609,7 +609,7 @@ async def company_tree(
  """
  logger.info(f"GET /api/companies/{company_name}/tree start")
 
- df = get_company_tree(company_name=company_name)
+ df = get_company_tree(company_domain=company_name)
 
  parent_company = df["parent_company_name"].tolist()[0]
 
@@ -620,7 +620,7 @@ async def company_tree(
  df[
  ~(parent_company == df["company_name"])
  & (company_name == df["company_name"])
- ]["domain"]
+ ]["company_domain"]
  .unique()
  .tolist()
  )
@@ -630,7 +630,7 @@ async def company_tree(
  ~(parent_company == df["company_name"])
  & (company_name != df["company_name"])
  ]
- .rename(columns={"domain": "domains"})
+ .rename(columns={"company_domain": "domains"})
  .groupby(["company_name"])["domains"]
  .apply(lambda x: list(x))
  .reset_index()
@@ -669,7 +669,7 @@ async def company_sdks(
  """
  logger.info(f"GET /api/companies/{company_name}/sdks start")
 
- df = get_company_sdks(company_name=company_name)
+ df = get_company_sdks(company_domain=company_name)
 
  mydict = CompanyPatternsDict(
  companies={

diff --git a/backend/dbcon/queries.py b/backend/dbcon/queries.py
@@ -254,7 +254,7 @@ def get_company_overview(company_domain: str) -> pd.DataFrame:
  df = pd.read_sql(
  QUERY_COMPANY_PARENT_OVERVIEW,
  DBCON.engine,
- params={"company": company_domain},
+ params={"company_domain": company_domain},
  )
  if df.empty:
  df = pd.read_sql(
@@ -267,24 +267,24 @@ def get_company_overview(company_domain: str) -> pd.DataFrame:
  return df
 
 
-def get_company_tree(company_name: str) -> pd.DataFrame:
+def get_company_tree(company_domain: str) -> pd.DataFrame:
  """Get a company tree with parent companies and domains."""
- logger.info(f"query company tree: {company_name=}")
+ logger.info(f"query company tree: {company_domain=}")
  df = pd.read_sql(
  QUERY_COMPANY_TREE,
  DBCON.engine,
- params={"company": company_name},
+ params={"company_domain": company_domain},
  )
  return df
 
 
-def get_company_sdks(company_name: str) -> pd.DataFrame:
+def get_company_sdks(company_domain: str) -> pd.DataFrame:
  """Get a company tree with parent companies and domains."""
- logger.info(f"query company sdks: {company_name=}")
+ logger.info(f"query company sdks: {company_domain=}")
  df = pd.read_sql(
  QUERY_COMPANY_SDKS,
  DBCON.engine,
- params={"company": company_name},
+ params={"company_domain": company_domain},
  )
  return df
 

diff --git a/backend/dbcon/sql/query_company_parent_overview.sql b/backend/dbcon/sql/query_company_parent_overview.sql
@@ -8,5 +8,5 @@ SELECT
 FROM
  adtech.companies_parent_app_counts
 WHERE company_domain = :company_domain
-GROUP BY company_domain, store, tag_source, app_category
+GROUP BY company_domain, company_name, store, tag_source, app_category
 ORDER BY app_count DESC;
diff --git a/backend/dbcon/sql/query_company_sdks.sql b/backend/dbcon/sql/query_company_sdks.sql
@@ -11,6 +11,18 @@ FROM
 LEFT JOIN adtech.companies AS cc
  ON
  c.parent_company_id = cc.id
+LEFT JOIN adtech.company_domain_mapping AS cdm
+ ON
+ c.id = cdm.company_id
+LEFT JOIN adtech.company_domain_mapping AS pcdm
+ ON
+ cc.id = pcdm.company_id
+LEFT JOIN ad_domains AS ad
+ ON
+ cdm.domain_id = ad.id
+LEFT JOIN ad_domains AS parad
+ ON
+ pcdm.domain_id = parad.id
 LEFT JOIN adtech.sdk_packages AS sp
  ON
  c.id = sp.company_id
@@ -20,5 +32,5 @@ LEFT JOIN adtech.sdk_paths AS sp2
  c.id = sp2.company_id
  OR cc.id = sp2.company_id
 WHERE
- c.name = :company
- OR cc.name = :company;
+ ad.domain = :company_domain
+ OR parad.domain = :company_domain
diff --git a/backend/dbcon/sql/query_company_tree.sql b/backend/dbcon/sql/query_company_tree.sql
@@ -1,23 +1,38 @@
-SELECT
- c.name AS company_name,
- ad.domain,
- COALESCE(
- cc.name,
- c.name,
- ad.domain
- ) AS parent_company_name
+WITH mytree AS (
+ SELECT
+ c.name AS company_name,
+ ad.domain AS company_domain,
+ COALESCE(
+ parad.domain,
+ ad.domain
+ ) AS parent_company_domain,
+ COALESCE(
+ cc.name,
+ c.name,
+ ad.domain
+ ) AS parent_company_name
+ FROM
+ adtech.companies AS c
+ LEFT JOIN adtech.companies AS cc
+ ON
+ c.parent_company_id = cc.id
+ LEFT JOIN adtech.company_domain_mapping AS cdm
+ ON
+ c.id = cdm.company_id
+ LEFT JOIN adtech.company_domain_mapping AS pcdm
+ ON
+ cc.id = pcdm.company_id
+ FULL OUTER JOIN ad_domains AS ad
+ ON
+ cdm.domain_id = ad.id
+ LEFT JOIN ad_domains AS parad
+ ON
+ pcdm.domain_id = parad.id
+)
+
+SELECT *
 FROM
- adtech.companies AS c
-LEFT JOIN adtech.companies AS cc
- ON
- c.parent_company_id = cc.id
-LEFT JOIN adtech.company_domain_mapping AS cdm
- ON
- c.id = cdm.company_id
-FULL OUTER JOIN ad_domains AS ad
- ON
- cdm.domain_id = ad.id
+ mytree
 WHERE
- c.name = :company
- OR cc.name = :company
- OR ad.domain ILIKE :company || '%';
+ company_domain = :company_domain
+ OR parent_company_domain = :company_domain;