Scenario:
a site (on older version) has a custom app xyz installed.
this site is to be restored on a new bench (without xyz) and migrated to recent version.
Problem:
`bench migrate` fails because xyz is not available.
`bench remove-from-installed-apps` might fail, because code is not compatible with the database.
db_host != the hostname part after @, hmm. so instead we'll pass
no-mariadb-socket flag to setup_database where we'll drop and create
users with host set to the TCP/IP socket instead of the UNIX socket.
Signed-off-by: Chinmay D. Pai <chinmaydpai@gmail.com>
* Allow setting db port on install
* Allow setting db port on install
* Allow full db config on creating site
* bug fix
* bug fix
* Escape character
* Update frappe/commands/site.py
allow only integer argument
Co-Authored-By: Chinmay Pai <chinmaydpai@gmail.com>
* Update frappe/database/db_manager.py
No need to escape integer variable
Co-Authored-By: Chinmay Pai <chinmaydpai@gmail.com>
* cast port to integer to surpress pymysql error
* cast port to string to surpress pymysql error
* Removed type casting
* [start] postgres
* [wip] started refactoring db_schema
* Add psycopg2 to requirements.txt
* Add support for Postgres SQL
- Separate frameworkSQL, database, schema, setup_db file for
mariaDB and postgres
- WIP
* Remove quotes from sql to make it compatible with postgres as well
* Moved some code from db_schema to database.py
* Move code from db_schema to schema.py
Add other required refactoring
* Add schema chages
* Remove redundant code in file
* Add invalid column name exception class to exceptions.py
* Add back tick in query wherever needed and replace ifnull with coalesce
* Update get_column_description code in database.py file
* Remove a print statement
* Add keys to get on_duplicate query
* Add bactick wherever necessary
- Remove db_schema.py file
* Remove DATE_SUB as it is incompatible with postgres
- Fix prepare_filter_condition
* Add backtick and quotes wherever necessary
- Move get_database_size to frappe.db namespace
- fix some left out bugs and errors
* Add code to create key and unique index
- added mysql and posgres in their respective database.py
* Add more bacticks in queries and fix some errors
- Pass keys to on_duplicate_update method
- Replace MONTH with EXTRACT function
- Remove DATEDIFF and CURDATE usage
* Cast state value to int in toggle_two_factor_auth
- since two_factor_auth has the datatype of Int
* Refactor
- Replace Timediff with normal arithmetic operator
- Add MAX_COLUMN_LENGTH
- Remove Redundant code
- Add regexp character constant
- Move create_help_table to database.py
- Add get_full_text_search_condition method
- Inherit MariaDBTable from DBTable
* Replace Database instance with get_db method
* Move db_manager to separate file
* Refactor
- Remove some unwanted code
- Separate alter table code for postgres and mysql
- Replace data_type with column_type in database.py
* Make fulltext search changes in global_search.py
* Add empty string check
* Add root_password to site config
* Create cli command for postgres console
* Move setup of help database to setup_db.py
* Add get_database_list method
* Fix exception handling
- Replace bad_field handler with missing_column handler
* Fix tests and sql queries
* Fix import error
* Fix typo db -> database
* Fix error with make_table in help.py
* Try test for postgres
* Remove pyhton 2.7 version to try postgres travis test
* Add test fixes
* Add db_type to the config of test_site_postgres
* Enable query debug to check the reason for travis fail
* Add backticks to check if the test passes
* Update travis.yml
- Add postgres addon
* Try appending 'd_' to hash for db_name
- since postgres does not support dbname starting with a number
* Try adding db_type for global help to make travis work
* Add print statements to debug travis failure
* Enable transaction and remove debug flag
* Fix help table creation query (postgres)
* Fix import issue
* Add some checks to prevent errors
- Some doctypes used to get called even before they are created
* Try fixes
* Update travis config
* Fix create index for help table
* Remove unused code
* Fix queries and update travis config
* Fix ifnull replace logic (regex)
* Add query fixes and code cleanup
* Fix typo
- get_column_description -> get_table_columns_description
* Fix tests
- Replace double quotes in query with single quote
* Replace psycopg2 with psycopg2-binary to avoid warnings
- http://initd.org/psycopg/docs/install.html#binary-install-from-pypi
* Add multisql api
* Add few multisql queries
* Remove print statements
* Remove get_fulltext_search_condition method and replace with multi query
* Remove text slicing in create user
* Set default for 'values' argument in multisql
* Fix incorrect queries and remove few debug flags
- Fix multisql bug
* Force delete user to fix test
- Fix Import error
- Fix incorrect query
* Fix query builder bug
* Fix bad query
* Fix query (minor)
* Convert boolean text to int since is_private has datatype of int
- Some query changes like removed double quotes
and replace with interpolated string to pass multiple
value pass in one of the query
* Extend database class from an object to support python 2
* Fix query
- Add quotes around value passed to the query for variable comparision
* Try setting host_name for each test site
- To avoid "RemoteDisconnected" error while testing data migration test
- Update travis.yml to add hosts
- Remove unwanted commit in setup_help_database
* Set site hostname to data migration connector (in test file)
- To connect the same site host
* Fix duplicate entry issue
- the problem is in naming series file.
In previous commits I unknowingly changed a part of a series query
due to which series were not getting reset
* Replace few sql queries with orm methods
* Fix codacy
* Fix 'Doctype Sessions not found' issue
* Fix bugs induced during codacy fixes
* Fix Notification Test
- Use ORM instead of raw sql
* Set Date fallback value to 0001-01-01
- 0000-00-00 is invalid date in Postgres
- 0001-01-01 works in both
* Fix date filter method
* Replace double quotes with single quote for literal value
* Remove print statement
* Replace double quotes with single
* Fix tests
- Replace few raw sql with ORM
* Separate query for postgres
- update_fields_to_fetch_query
* Fix tests
- replace locate with strpos for postgres
* Fix tests
- Skip test for datediff
- convert bytes to str in escape method
* Remove TestBot
* Skip fieldname extraction
* Replace docshare raw sql with ORM
* Fix typo
* Fix ancestor query test
* Fix test data migration
* Remove hardcoded hostname
* Add default option and option list for db_type
* Remove frappe.async module
* Remove a debug flag from test
* Fix codacy
* fix import issue
* Convert classmethod to static method
* Convert few instance methods to static methods
* Remove some unused imports
* Fix codacy
- Add exception type
- Replace few instance methods with static methods
- Remove unsued import
* Fix codacy
* Remove unused code
* Remove some unused codes
- Convert some instance methods to static function
* Fix a issue with query modification
* Fix add_index query
* Fix query
* Fix update_auth patch
* Fix a issue with exception handling
* Add try catch to a reload_doc
* Add try-catch to file_manager_hook patch
* import update_gravatar to set_user_gravatar patch
* Undo all the wrong patch fixes
* Fix db_setup code 😪
- previously it was not restoring db from source SQL
which is why few old patched were breaking
(because they were getting different schema structure)
* Fix typo !
* Fix exception(is_missing_column) handling
* Add deleted code
- This code is only used in a erpnext patch.
Can be moved to that patch file
* Fix codacy
* Replace a mariadb specific function in a query used in validate_series
* Remove a debug flag
* Revert changes (rename_parent_and_child)
* Fix validate_one_root method
* Fix date format issue
* Fix codacy
- Disable a pylint for variable argument warning
- Convert an instance method to static method
* Add bandit.yml
The Codacy seems to use Bandit which generates
warning for every subprocess import and its usage during pytest
Since we have carefully used subprocess (avoided user input),
warnings needs to be avoided.
This can be removed if we have any alternative for subprocess usage.
* Skip start_process_with_partial_path check
* Fix typo
* Add python 2.7 test
* Move python versions in travis.yml
* Add python versions to jobs
* Overwrite python version inheritance for postgres in travis.yml
* Add quotes around python version in .travis.yml
* Add quotes around the name of the job
* Try a travis fix
* Try .travis.yml fix
* Import missing subprocess
* Refactor travis.yml
* Refactor travis.yml
- move install and tests commands to separate files
- Use matrix to build combination of python version and db type
* Make install.sh and run-tests.sh executable
* Add sudo required to travis.yml to allow sudo cmmands in shell files
* Load nvm
* Remove verbose flag from scripts
* Remove command-trace-print flag
* Change to build dir in before script
* Add absolute path for scripts
* Fix tests
* Fix typo
* Fix codacy
- fixes - "echo won't expand escape sequences." warning
* Append (_) underscore instead of 'd' for db_name
* Remove printf and use mysql execute flag
* Moving after_install to occur after fixtures and customizations install
"after_install" was being called prior to fixtures that a module depends on gets installed, which makes after_install incapable of working with those doctypes. This is particularly frustrating because there is not an "after_install_with_fixtures_and_customizations" hook...
* fixing trailing whitespace
* Fix for notifications using "from_module".
Notifications using "from_module" designation previously did not work. There's two reasons for this. First, the calculation for summing notifications was exclusively choosing the notifications counts from doctype, and not using the total sum to include modules. Second, the default "_id" for a module was being automatically set to the "link" of a module. This is an issue because module links can actually contain JS (which allow them to open there pages in the same window, for example). Because of this, they do not make for good identifiers. This fix uses the toLowerCase of the module_name as the default module _id.
* fixing double declaration of count
* contending with "core is already defined" codacy error.
* Adding after_sync hook per @rmehta request.
* trying to force travis-CI to rebuild
* adds new exception to be raised when any improper database configuration is detected
* changes behavior of `check_if_ready_for_barracuda` to raise ImproperDBConfigurationError instead of sys.exit`
* refactors `drop_site` to use a new exactly identical `_drop_site` function.
The reason for this is because the original `drop_site` function is decorated
and cannot be undecorated without nasty hacks. Breaking the function this way
allows me to make use of the `drop-site` logic easily.
* catches the ImproperDBConfigurationError raised from `check_if_ready_for_barracuda` function to drop all the artifacts of the failed new `Site`
* creates new function `print_db_config` to print out ERPNext required database settings to console
* refactors `check_if_ready_for_barracuda` to make use of `print_db_config`
* adds a more explicit explanation before showing the `expected_config_for_barracuda` text
* adds MariaDB default configuration to documentation
* [list_settings] save last_view, kanban_board, gantt_mode
* listview.js cleanup
* image_view refactor
* image_view done
* wip
* [photoswipe] wip
* show view for calendar, kanban even if no results
* [gantt_view] refactor into separate file
* [imageview] 3 column border fix
* [imageview] gallery working
* delete old libs
* indentation to tabs
* [gantt] update lib, custom popup html
* custom fontawesome checkbox
* reset gantt state when not permitted
* checkbox styling fix
* working commit
* image, calendar, gantt view working
* more refactoring, kanban view
* minor
* removed old files
* user settings improved
* filters and sort selector de-coupling
* wip
* [imageview] white pswp background
* kanban filters saving fixed
* fixed reportview
* minor
* removed listing.js
* minor fixes and cleanup
* patch for UserSettings table
* patch fix
* [start] global search frappe/erpnext#6674
* [fix] setup before running test
* [start] global search frappe/erpnext#6674
* Display result as rudimentary list, rebuild old doctypes
* Media view, child tables, delete document updates, searchable fields
* More results UI
* Code clean up
* remove msgprint from document.py to resolve merge conflict
* Modularization stage 1, get show more to work with it
* Dedicated modal Search bar works, some clean up needed
* Can't data-dismiss on links, Bootstrap issue, use hashchange
* Accomodate missing field content syndrome
* Search in boolean mode, make GS default in awesome bar, fix double modal bug and cleanup
* Add in Meta
* Add in customize form
* Modularise Global Search
* Search object
* Commonify Search UI: Stage I
* II: save list state, UI, default condensed view, refactor
* Fix SQL bug, Refactor awesome bar, Fix unicode bug, add nav results
* Refactor using separate search objects, some async issues
* Fix async flow
* Fix preceding more list bug
* UI additions
* another async fix, back link
* Help: Stage I
* Help: Stage II
* Background jobs, fix route options bug
* Fix GS syncing on install
* Add GS options in awesome bar: test
* Input now remembers search type state
* More UI updates
* Add description for GS results in awesome bar
* Fix help modal bug
* Fix: not commit during install
* Test cases, some fixes
* Update in_test flag in enqueue
* Disable GS sync when not install_db
* Add flag check
* Disable field in child tables
* Cleanups
* Create table fix
* Fix redis exception, remove commit enqueue, add gs in migrate
* Fix tests
* Single enqueue
* cleanups
* Fix tests
* Fix event test
* Fix duplication, search as first option
* Add show name in global search
* fix event tests and desk.less
* Fix communication.json
* [fixes] wip
* [fix] tests
* [minor] for tests
* [minor] for tests
* [minor] for tests
* [minor] for tests
