Update nutconf and augeas lenses with support for NUT keywords added between releases 2.6.5 and 2.8.2

[jimklimov]

Closes: #2294

Note: per #657 there is more work to improve flexibility of augeas lenses - simple cases should work, but not all combos that NUT own parser supports.

Note: It seems that per initial codebase import (from 2013-ish branch to NUT v2.8.2) the sprawl of C++ header/class/test code is an internal detail of nutconf CLI tool - not a library delivered for external consumer linking. As such, there are no concerns currently about API versioning, backwards compatibility, variable/method name changes, etc. which would impact any code not visible from NUT repo itself and its regular builds. This frees up hands for refactoring like this, but eventually it could be useful to document and release the library for NUT configuration management... maybe? Or would the CLI suffice for external integrations?

[jimklimov]

Hmm, this typeid equality check seems to exceed the capabilities of older compilers (centos7 - https://ci.networkupstools.org/job/nut/job/nut/job/PR-2425/6/artifact/.ci.MD5_f52f7a4bde4e35935a6dd5c50b706166.build.log.gz or Ubuntu Trusty https://ci.networkupstools.org/job/nut/job/nut/job/PR-2425/6/artifact/.ci.MD5_6151b77274dd5261710fbd7a7f634ff8.build.log.gz in a recent CI run):

nutconf.cpp:59:16: error: invalid operands to binary expression ('const std::type_info' and 'const std::type_info')
        if (typeid(T) == typeid(bool)) {
            ~~~~~~~~~ ^  ~~~~~~~~~~~~
/usr/bin/../lib/gcc/x86_64-redhat-linux/4.8.5/../../../../include/c++/4.8.5/bits/stl_pair.h:214:5: note: candidate template ignored: could not match 'pair<type-parameter-0-0, type-parameter-0-1>' against 'const std::type_info'
    operator==(const pair<_T1, _T2>& __x, const pair<_T1, _T2>& __y)
    ^

nutwriter.cpp:296:2: error: invalid operands to binary expression ('const std::type_info' and 'const std::type_info')
        SHELL_CONFIG_DIRECTIVEX("ALLOW_NO_DEVICE",              bool,                   config.allowNoDevice,   false);
        ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
nutwriter.cpp:94:24: note: expanded from macro 'SHELL_CONFIG_DIRECTIVEX'
                        if (typeid(arg_val) == typeid(bool&)) { \
                            ~~~~~~~~~~~~~~~ ^  ~~~~~~~~~~~~~

Wondering if it is not beyond salvation... is there some equals() method? Or some missed include...

[jimklimov]

Not all systems were created equal... How does augeas decide this, when other regex parsers fail to match the beast to slash. Would escaped \\t\\r\\n of sorts (see last chars) help?

Syntax error in lens definition
./nutupsconf.aug:296.0-.99:Failed to compile ups_entry
/usr/local/share/augeas/lenses/dist/inifile.aug:269.34-.40:exception: The key regexp /((((((((((((chroot)|(driverpath))|(maxstartdelay))|(maxretry))|(nowait))|(retrydelay))|(pollinterval))|(synchronous))|(user))|(group))|(debug_min))|(((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((driver)|(port))|(sdorder))|(desc))|(nolock))|(ignorelb))|(maxstartdelay))|(synchronous))|(user))|(group))|(debug_min))|(BYPASS))|(CHRG_addr))|(CHRG_noro))|(CHRG_regtype))|(CP))|(CS))|(DISCHRG_addr))|(DISCHRG_noro))|(DISCHRG_regtype))|(FSD_addr))|(FSD_noro))|(FSD_pulse_duration))|(FSD_regtype))|(HB_addr))|(HB_noro))|(HB_regtype))|(ID))|(LB))|(LB_addr))|(LB_noro))|(LB_regtype))|(LowBatt))|(OB_addr))|(OB_noro))|(OB_regtype))|(OL))|(OL_addr))|(OL_noro))|(OL_regtype))|(OffDelay))|(OnDelay))|(RB))|(RB_addr))|(RB_noro))|(RB_regtype))|(SD))|(advanced_eco_mode))|(advorder))|(alarm_control))|(allow_duplicates))|(authPassword))|(authProtocol))|(authtype))|(awd))|(batteryPercentage))|(battery_alarm))|(battery_max))|(battery_min))|(battery_number))|(battery_open_status_check))|(battery_voltage_reports_one_pack))|(battext))|(battvoltmult))|(baud_rate))|(baudrate))|(bus))|(busport))|(bypass_alarm))|(bypass_forbidding))|(bypass_when_off))|(cable))|(cablepower))|(chargetime))|(community))|(constant_phase_angle))|(converter_mode))|(cshdelay))|(databits))|(daysoff))|(daysweek))|(dev_slave_id))|(device))|(device_mfr))|(device_model))|(disable_fix_report_desc))|(do_convert_deci))|(dumbterm))|(eco_mode))|(explore))|(fake_lowbatt))|(fault_1))|(fault_2))|(fault_3))|(fault_4))|(fault_5))|(flash))|(frequency))|(fruid))|(full_update))|(hb))|(houroff))|(houron))|(i2c_address))|(idleload))|(ignoresab))|(input_fault_voltage))|(input_timeout))|(interruptonly))|(interruptsize))|(langid_fix))|(lb))|(limited_runtime_on_battery))|(linevoltage))|(load\.off))|(load\.on))|(load\.status))|(loadPercentage))|(localcalculation))|(login))|(lowbatt))|(manufacturer))|(max_bypass_freq))|(max_bypass_volt))|(max_load))|(max_polls_without_data))|(maxreport))|(methodOfFlowControl))|(mfr))|(mibs))|(min_bypass_freq))|(min_bypass_volt))|(mincharge))|(minruntime))|(mod_byte_to_s))|(mod_byte_to_us))|(mod_resp_to_s))|(mod_resp_to_us))|(model))|(modelname))|(nobt))|(nohang))|(nombattvolt))|(nominal_cell_voltage))|(norating))|(noscanlangid))|(notification))|(notransferoids))|(novendor))|(nowarn_noimp))|(numOfBytesFromUPS))|(number_of_battery_cells))|(offdelay))|(oldmac))|(ondelay))|(onlinedischarge))|(onlinedischarge_calibration))|(onlinedischarge_log_throttle_hovercharge))|(onlinedischarge_log_throttle_sec))|(onlinedischarge_onbattery))|(output_pace))|(output_phase_angle))|(output_voltages))|(parity))|(password))|(pins_shutdown_mode))|(pollfreq))|(pollonly))|(porttype))|(powerup))|(prefix))|(prgshut))|(privPassword))|(privProtocol))|(product))|(productid))|(protocol))|(rebootdelay))|(recharge_time))|(repeater_disable_strict_start))|(reset_to_default))|(response_timeout_ms))|(rio_slave_id))|(rules))|(runtime_full))|(runtime_half))|(runtimecal))|(sdtime))|(sdtype))|(secLevel))|(secName))|(semistaticfreq))|(sensorid))|(ser_baud_rate))|(ser_data_bit))|(ser_parity))|(ser_stop_bit))|(serial))|(serialnumber))|(series))|(shutdownArguments))|(shutdown_delay))|(shutdown_duration))|(shutdown_timer))|(silent))|(site_fault_detection))|(slave_address))|(slaveid))|(snmp_retries))|(snmp_timeout))|(snmp_version))|(startdelay))|(status_only))|(stayoff))|(stopbits))|(subdriver))|(subscribe))|(symmetrathreephase))|(testing))|(testtime))|(timeout))|(ttymode))|(type))|(ups\.delay\.shutdown))|(ups\.delay\.start))|(upsid))|(upstype))|(usb_config_index))|(usb_hid_desc_index))|(usb_hid_ep_in))|(usb_hid_ep_out))|(usb_hid_rep_index))|(usb_set_altinterface))|(usd))|(use_crlf))|(use_pre_lf))|(username))|(validationSequence))|(vendor))|(vendorid))|(voltage))|(wait))|(waitbeforereconnect))|(work_range_type))|(wugrace)))|((default|override)\.[^:=#	

 ]+)/ matches a '/' which is used to separate nodes.

[jimklimov]

Hm... seems the universal augeas parser for handling configs portably is not portable itself as an engine. Same NUT lens and test files fail or pass with different messages, sometimes regardless of augeas versions (one is passing, but older/newer ones fail).

[jimklimov]

I guess I'll have to disable the augeas self-test by default again, let the scope of issue #657 deal with it. Was inclined to cut off by release version, but augeas-1.12.0 on Debian and FreeBSD behave differently vs. same NUT test codebase...

First will try to separate a "flaky" test with default.* pattern handling into a non-default file to use in manual tests but not run blindly everywhere.

This passes on various linuxes but fails on FreeBSD and OpenBSD workers (used to fail similarly on Ubuntu but then somehow the syntax got wedged to augeas' liking):

~/nut/scripts/augeas$ ./gen-nutupsconf-aug.py ; make ; augparse --trace -I ./ ./tests/test_nut_flaky.aug && echo PASSED
Module ./tests/test_nut_flaky.aug loaded
Module ./nutupsconf.aug loaded
Module /usr/local/share/augeas/lenses/dist/inifile.aug loaded
Module /usr/local/share/augeas/lenses/dist/util.aug loaded
Module /usr/local/share/augeas/lenses/dist/sep.aug loaded
Module /usr/local/share/augeas/lenses/dist/rx.aug loaded
Module /usr/local/share/augeas/lenses/dist/quote.aug loaded
Module /usr/local/share/augeas/lenses/dist/build.aug loaded
./tests/test_nut_flaky.aug:21.0-27.47:exception thrown in test
./tests/test_nut_flaky.aug:21.5-.37:exception: Get did not match entire input
    Lens: /usr/local/share/augeas/lenses/dist/inifile.aug:497.25-.43:
    Error encountered at 6:0 (71 characters into string)
    <\n\tdesc = "Dummy UPS Driver"\n|=|\tdefault.battery.voltage.hig>

    Tree generated so far:
    /testups
/testups/driver = "dummy-ups"
/testups/port = "auto"
/testups/desc = "Dummy UPS Driver"


Syntax error in lens definition
Failed to load ./tests/test_nut_flaky.aug

[jimklimov]

Yeah, so after all - the quick way forward with this was to keep augeas self-check ability in cold storage as far as regular builds are concerned. It should be a bit more facilitated for developers now though.